Upload
haduong
View
219
Download
0
Embed Size (px)
Citation preview
Independent Evaluation of
Budget Support in Mozambique
Final Report Volume I 2014
________
Evaluation jointly managed by the European
Commission, the Ministry of Planning and Development of Mozambique, and the evaluation
departments of Ireland, Germany, Belgium, Italy, Finland, the Netherlands and France
Consortium of
ADE, ITAD and COWI
Consortium leader: ADE s.a
Contact Person: Jean-Marie Wathelet
Contract No EVA 2011/Lot 3
This evaluation was commissioned by the Evaluation Unit of the
Directorate General for Development and Cooperation – EuropeAid
(European Commission)
It was jointly managed by the European Commission, the Ministry of
Planning and Development of Mozambique, and the evaluation
departments of Ireland, Germany, Belgium, Italy, Finland, the
Netherlands and France
The opinions expressed in this document represent the authors’ point of view which are not necessarily shared by the European Commission or by
the authorities of the concerned countries.
This report has been prepared by
12 English Business Park
English Close, Hove, East Sussex, BN3 7ET United Kingdom
T: +44 1273 765250 Web: itad.com
In collaboration with
and
Image © Mozambique, Chokwe, Lhate village: Photo credit: ILRI
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | iii
Evaluation Team Andrew Lawson Team Leader
Ann Bartholomew Deputy Team Leader
Mariam Bibi PFM Specialist
Hermes Sueia Agriculture Specialist
Padil Salimo Governance Specialist
Tim Cammack Health Specialist
Muriel Visser Education Specialist
Andy McKay Senior Econometrician
Giulia Mascagni Econometrician
Edgar Salgado Econometrician
Gonzalo Contreras Economist/Statistician
Giorgia Giambi Research Assistant
Evildo Semo Research Assistant
Tim Ruffer Quality Assurance/Project Manager
Karolyn Thunnissen Quality Assurance
Jodie Ellis Project Officer
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | iv
Table of Contents
FIGURES ................................................................................................................................................................................... VI TABLES ................................................................................................................................................................................... VII TEXT BOXES .......................................................................................................................................................................... VIII ACRONYMS AND ABBREVIATIONS ......................................................................................................................................... IX
EXECUTIVE SUMMARY ................................................................................................................................................... XII
1. INTRODUCTION ....................................................................................................................................................... 20 1.1. SCOPE AND OBJECTIVES OF THE EVALUATION ....................................................................................................20 1.2. THE EVALUATION FRAMEWORK AND PROCESS ....................................................................................................22 1.3. EVALUATION QUESTIONS ........................................................................................................................................25 1.4. STRUCTURE OF THE REPORT...................................................................................................................................28
2. THE CONTEXT FOR BUDGET SUPPORT IN MOZAMBIQUE: HELP OR HINDRANCE? ........................ 30 2.1. THE EVOLVING SOCIO-ECONOMIC CONTEXT OF MOZAMBIQUE ..........................................................................30 2.2. THE EVOLVING FRAMEWORK OF DEMOCRATIC ACCOUNTABILITY .......................................................................35 2.3. FLUCTUATIONS IN THE POLICIES, PRIORITIES AND ATTITUDES OF THE DEVELOPMENT PARTNERS ................41 2.4. IMPLICATIONS FOR THE EVALUATION AND FOR FUTURE BUDGET SUPPORT .....................................................42
3. PROVIDING THE MEANS TO IMPLEMENT POLICY: BUDGET SUPPORT INPUTS AND DIRECT OUTPUTS ............................................................................................................................................................................. 45
3.1. SCALE & PREDICTABILITY OF BUDGET SUPPORT, AND ITS EFFECTS ON AGGREGATE AID FLOWS .................46 3.2. INFLUENCE OF BUDGET SUPPORT ON POLICY DIALOGUE PROCESSES ...............................................................55 3.3. CONTRIBUTIONS TO CAPACITY BUILDING ..............................................................................................................70 3.4. ADAPTING THE BUDGET SUPPORT DESIGN TO THE CHANGING CONTEXT ..........................................................72
4. FACILITATING IMPROVEMENTS IN GOVERNMENT POLICIES: INDUCED OUTPUTS OF BUDGET SUPPORT ............................................................................................................................................................................. 82
4.1. MACROECONOMIC MANAGEMENT & REVENUE GENERATION ..............................................................................82 4.2. THE QUALITY OF PUBLIC FINANCIAL MANAGEMENT ...........................................................................................91 4.3. CHANGES IN THE COMPOSITION OF PUBLIC SPENDING .......................................................................................99 4.4. CHANGES INDUCED IN SECTORAL POLICIES ....................................................................................................... 105 4.5. GOVERNANCE AND THE FIGHT AGAINST CORRUPTION ..................................................................................... 117
5. GROWTH & POVERTY REDUCTION ............................................................................................................... 129 5.1. GROWTH DATA AND TRENDS BY ECONOMIC SECTOR AND PROVINCE ............................................................. 130 5.2. TRENDS IN CONSUMPTION POVERTY BY PROVINCE .......................................................................................... 133 5.3. CONCLUSIONS ON INCOME POVERTY IN MOZAMBIQUE ................................................................................... 136 5.4. TRENDS IN NON-INCOME POVERTY .................................................................................................................... 137 5.5. CONCLUSIONS ON GROWTH & POVERTY IN MOZAMBIQUE ............................................................................ 141
6. AGRICULTURAL SECTOR OUTCOMES AND THEIR DETERMINANTS .................................................. 142 6.1. TRENDS IN AGRICULTURAL PERFORMANCE ........................................................................................................ 142 6.2. REGRESSION ANALYSIS: PRODUCTIVITY AND INPUT USE .................................................................................. 146 6.3. CONCLUSIONS ON DETERMINANTS OF AGRICULTURAL PRODUCTIVITY ........................................................... 149
7. EDUCATION SECTOR OUTCOMES AND THEIR DETERMINANTS ......................................................... 150 7.1. OVERVIEW OF ENROLMENT & OTHER KEY DATA ............................................................................................... 150 7.2. ANALYSIS OF THE DETERMINANTS OF EDUCATIONAL OUTCOMES ................................................................... 156 7.3. CONCLUSIONS: KEY DETERMINANTS OF EDUCATION OUTCOMES .................................................................... 160
8. CONCLUSIONS AND RECOMMENDATIONS ................................................................................................ 162 8.1. OVERALL FINDINGS .............................................................................................................................................. 162 8.2. STEP THREE: CONCLUSIONS ON IMPACT OF BUDGET SUPPORT ..................................................................... 170 8.3. RECOMMENDATIONS ............................................................................................................................................ 176
BIBLIOGRAPHY .............................................................................................................................................................. 182
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | v
ANNEXES (IN VOLUME 2)
ANNEX 1: SUMMARY OF BUDGET SUPPORT OPERATIONS EVALUATED ANNEX 2: SUMMARY OF RESPONSES TO EVALUATION QUESTIONS AND CORRESPONDING DATA SOURCES ANNEX 3: DATA TABLES ON AID, FISCAL FLOWS AND PUBLIC EXPENDITURES ANNEX 4: BIBLIOGRAPHY ANNEX 5: PEOPLE MET
Figures
Figure 1: The 5 levels and the 3 Steps of the Evaluation methodology ...................................................22
Figure 2: Budget Support Intervention Logic in Mozambique and proposed Evaluation Questions .....................................................................................................................................................................................................23
Figure 3: Map of Mozambique .......................................................................................................................................28
Figure 4: Real GDP for Mozambique in comparison with other ‘frontier economies’ .........................30
Figure 5: Voter turn-out and invalid votes in Mozambique & neighbouring countries ......................37
Figure 6: Importance in public spending of Budget Support and other aid modalities ....................46
Figure 7: Actual annual disbursements as a percentage of the planned disbursements: Overall predictability of aid (left); Individual detail for each modality (right) .......................................................47
Figure 8: The “lost” expenditure due to low execution rates of Common Basket Funds and externally financed Projects ..........................................................................................................................................48
Figure 9: Budget support disbursements “front-loaded” in the first two quarters ............................48
Figure 10: Differences between scheduled & actual GBS disbursements by Quarter (Millions of MZM) ..........................................................................................................................................................................................49
Figure 11: Results of PAPs’ indicators within the PAF over 2005 – 2012 .............................................51
Figure 12: Aggregate assessment of PAPs’ performance indicators, 2005-212 .................................52
Figure 13: Budget Support Coordination and Dialogue Mechanisms .........................................................55
Figure 14: Performance of Government PAF indicators, 2005 - 2012 ......................................................59
Figure 15: Performance of Government PAF indicators by sector/ theme ..............................................59
Figure 16: Assessment of Performance to determine EU MDG tranche for 2012-2014 ..................76
Figure 17: Revenue Collections 2006-2012 (% of GDP) ..................................................................................82
Figure 18: Mozambique : Coal and LNG Contribution to Fiscal Revenue, 2011-2032 .......................83
Figure 19: Consumer Price Index 2000-2012, compared with SSA and frontier economies ..........86
Figure 20: Global Competitiveness, 2012-2013: Mozambique & neighbouring countries ...............86
Figure 21: Comparison of PEFA Scores for Mozambique - 2006, 2008 and 2010 ..............................95
Figure 22: The contribution of Budget Support to the funding of discretionary spending ........... 100
Figure 23: Actual allocations to priority sectors compared with estimated allocations of Domestic Revenue, showing the funding gap filled by Budget Support (MZM Trillions) ................ 102
Figure 24: Estimates of Per Capita Health spending by Province, 2010 in Meticais ....................... 112
Figure 25: Allocations to Tribunal Administrativo compared with estimated allocation of Domestic Revenue, showing the funding gap filled by Budget Support (MZM Trillions) ................ 122
Figure 26: Allocations to Good Governance compared with estimated allocation of Domestic Revenue, showing the funding gap filled by Budget Support (MZM Trillions) ..................................... 126
Figure 27: Good Governance indicators in the PAF (left) and overall performance (right) .......... 126
Figure 28: GDP growth (constant prices) .............................................................................................................. 129
Figure 29: Growth in selected sectors .................................................................................................................... 129
Figure 30: Growth incidence curve at the national level, 2002/03 – 2008/09 .................................. 134
Figure 31: Trends in median agricultural productivity, 2002-08 .............................................................. 142
Figure 32: Regional pattern in Primary and Secondary Enrolment rates, 2008/09 ......................... 151
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | vi
Figure 33: Primary and Secondary Net Enrolment rates by sex and by year, 1997 - 2009 ........ 151
Figure 34: Extent of training of teachers in Primary Education, 2004 - 2012................................... 153
Figure 35: Index of availability of books per pupil in Primary schools, 2004=1.00 ....................... 154
Figure 36: Classroom construction material; Primary Schools, 2004-2012 ........................................ 154
Figure 37: Step Three – Summary of analysis of aggregate impact of Budget Support in Mozambique ........................................................................................................................................................................ 170
Figure 38: Diagrammatic Overview of Budget Support impacts and transmission mechanisms in Mozambique ........................................................................................................................................................................ 171
Tables
Table 1: Criteria for ranking of quality of Evidence for each Evaluation Question .............................24
Table 2: Economic & Social Indicators for Mozambique and Tanzania, 2005 and 2012 .................31
Table 3: Parliamentary Election results 1994 – 2009 (% of valid votes cast) .....................................35
Table 4: Voter turn-out and invalid votes cast in Mozambique & neighbouring countries ..............37
Table 5: Indicators of Political Openness and Perceptions of Corruption, 2005 & 2012 .................38
Table 6: Worldwide Governance Indicators for Voice & Accountability, 2006 & 2012 .....................39
Table 7: Budget Support Disbursements by Development Partner, 2004 - 2012 ................................45
Table 8: Disbursements of Budget Support and their significance .............................................................46
Table 9: Trends reported in use of Country systems: Paris Declaration Monitoring Surveys and indicators in PAPs’ Performance Assessment Framework ..............................................................................51
Table 10: Trends in PAPs performance indicators related to aid transaction costs ...........................54
Table 11: Pillars & Working Groups for PARPA/ Budget Support monitoring, 2004-2009 ..............56
Table 12: PARP/ Budget Support Working group structure 2011-2014 ....................................................57
Table 13: Total numbers of PAF indicators per year by Sector/ Thematic Area ...................................58
Table 14: Variable tranche disbursements, 2012 - 2014 ................................................................................73
Table 15: Minerals Sector Fiscal Frameworks for Mozambique & competing countries ..................83
Table 16: Overview of Central Government Fiscal Operations 2005-2012 (% of GDP) ...................84
Table 17: Real effective exchange rate and Budget Support receipts (2005-2011): .......................88
Table 18: Summary of PFM annual performance assessments in the Annual Reviews ....................91
Table 19: Key Events & Developments in PFM (including the IFMIS) .........................................................92
Table 20: Composition of Budget Funding (excluding non-concessional borrowing) ....................... 101
Table 21: Evolution of Spending within Priority Sectors during evaluation period ......................... 101
Table 22: Government Budget Allocations to the Agricultural Sector, 2005-2012 ......................... 106
Table 23: Aggregate Health spending by source of funds, 2005 and 2011 ........................................ 114
Table 24: Legislative scrutiny of the Budget - PEFA scores, 2006 - 2011 .......................................... 117
Table 25: Legislative Scrutiny of Audit Reports - PEFA scores, 2006 - 2011 ..................................... 118
Table 26: Scope, nature & follow-up of External Audit - PEFA scores, 2006 - 2011 .................... 120
Table 27: Audit Performance of the Tribunal Administrativo, 2003 -2013 ......................................... 121
Table 28 : Processing of Corruption Cases & Cases of Misuse of State Funds or Property: 2005-2012 ....................................................................................................................................................................................... 125
Table 29: Performance and relevance of PAF indicators for Governance, 2005 -2012 ................ 127
Table 30: Average growth in macro-sectors at the national level ........................................................... 130
Table 31: Mozambique - Average growth rates by province ....................................................................... 131
Table 32: Consumption, GDP and agriculture growth between survey years ..................................... 131
Table 33: Percentage Changes in poverty headcount (P0), poverty gap (P1) and squared poverty gap (P2), 1996/97 – 2008/ 09 ................................................................................................................................... 133
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | vii
Table 34: Growth-redistribution decomposition of changes in poverty from 2002/03 to 2008/09 .................................................................................................................................................................................................. 134
Table 35: Rate of pro-poor growth by Province, 2002/03 – 2008/09 .................................................... 135
Table 36: Percentage of households owning durable goods, 2002/ 03 and 2008/ 09 ................... 136
Table 37: Percentage of Households with Housing, sanitation and electricity ................................... 137
Table 38: Literacy rates (percentage of 15-30 age group) and Education (percentage of children attending school at right age), 2002/ 03 and 2008/ 09 ............................................................................... 138
Table 39: Child malnutrition (% of children two standard deviations from WHO international reference value) ................................................................................................................................................................ 138
Table 40: Median Area cultivated in Hectares .................................................................................................... 143
Table 41: Percentage of farmers using different inputs ............................................................................... 144
Table 42: Proportion of output sold, by crop type and year ........................................................................ 144
Table 43: Use of improved seeds and fertilizer and impact on productivity ....................................... 145
Table 44: Productivity and use of inputs by productivity percentile ....................................................... 145
Table 45: Regression Analysis for determinants of productivity at the farm level ......................... 146
Table 46: Regression Analysis of determinants of Productivity and Input use at the Community level ........................................................................................................................................................................................ 147
Table 47: Quantile regression for correlates of total productivity in crop cultivation ................... 148
Table 48: Key indicators for primary and secondary education in Mozambique, 2004-12 .......... 152
Table 49: Results for absolute enrolment ............................................................................................................ 156
Table 50: Results for Progression Rates ............................................................................................................... 157
Table 51: Results for Repetition ................................................................................................................................ 158
Table 52: Results for approval into Secondary School................................................................................... 159
Text Boxes
Box 1 Evaluation Questions ..........................................................................................................................................26
Box 2: Stakeholder’s feedback on the Incentive effects of Variable Tranches ...................................75
Box 3: Why PROSAUDE is not SBS and how GBS/ SBS could be diferent ............................................. 174
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | viii
This Report and Annexes (Volume 2) will be published on the EuropeAid website and
will be accessible at the following page:
http://ec.europa.eu/europeaid/how/evaluation/evaluation_reports/reports_by_year_en.ht
m
It will also be available on the Itad website at the following address: http://itad.com/knowledge-and-resources/reports/
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | ix
Acronyms and Abbreviations
ADE Apoio Direito as Escolas
Direct Grants to Schools
AFDB African Development Bank
APRM Africa Peer Review Mechanism
AR National Assembly
Assembleia da República
ATM Autoridade Tributária de Moçambique
Mozambique Tax Authority
CAADP Comprehensive Africa Agriculture Development Programme
CEDSIF Centro de Desenvolvimento de Sistemas de Informação de Finanças
Centre for Development of IT Finance Systems
CESC Centro de Aprendizagem e Capacitação da Sociedade Civil
Centre for Civil Society Learning & Capacity-building
CGE Conta Geral do Estado
General State Accounts
CIP Centro de Integridade Publica
Centre for Public Integrity
CPD Centro de Processamento de Dados
Data Processing Centre
CPO Comissão Parlamentar do Plano e Orçamento
Parliamentary Commission for Planning & the Budget
CRS Creditor Reporting System (OECD-DAC)
CUT Conta Única do Tesouro
Single Treasury Account
DAC Development Assistance Committee (of OECD)
Comité de Assistência ao Desenvolvimento (da OCDE)
DAF Directorate of Administration and Finance (in line ministries)
Direcção de Administração e Finanças
DNCP Direcção Nacional de Contabilidade Pública
National Directorate of Public Accounts
DNIC Direcção Nacional de Investimento e Cooperação
National Directorate of Investment and Cooperation
DNO Direcção Nacional do Orçamento
National Budget Directorate
DNT Direcção Nacional do Tesouro
National Treasury Directorate
DPs Development Partners
EC European Commission
Comissão Europeia
EMATUM Empresa Moçambicana de Atum
Mozambican Tuna Company
EQs Evaluation Questions
FASE Fundo de Apoio ao Sector de Educação
Education Sector Support Fund
FDI Foreign Direct Investment
FRELIMO Frente de Libertação de Moçambique
Mozambique Liberation Front
G-19 Group of General Budget Support Donors
Grupo de Doadores de Apoio Directo ao Orçamento
GBS General Budget Support
Apoio Geral ao Orçamento
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | x
GCCC Gabinete Central de Combate a Corrupção
Central Office for Combating of Corruption
GDP Gross Domestic Product
GFS Government Financial Statistics
Estatísticas Financeiras do Governo
GoM Government of Mozambique
Governo de Moçambique
IAF Inquérito aos Agregados Familiares
Household Survey
IESE Instituto de Estudos Sociais e Económicos
Institute of Social and Economic Studies
IMF International Monetary Fund
Fundo Monetário Internacional
IOF Inquérito ao Orçamento Familiar
Household Budget Survey
JAR Joint Annual Review
Revisão Conjunta
LNG Liquefied Natural Gas
MDA Ministries, Departments and Agencies (Budget-holding entities)
Ministérios, Departamentos e Instituições
MDM Democratic Movement of Mozambique (political party)
Movimento Democrático de Moçambique
MEC Ministry of Education and Culture
Ministério de Educação e Cultura
MISAU Ministry of Health
Ministério da Saúde
MoF Ministry of Finance
Ministério das Finanças
MoU Memorandum of Understanding
Memorando de Entendimento
MPD Ministry of Plan and Development
Ministério do Plano e Desenvolvimento
MTFF Medium-Term Fiscal Framework
Cenário Fiscal de Médio Prazo
ODAMOZ Official Development Assistance – Mozambique
(Database in MPD)
OECD Organisation for Economic Co-operation and Development
Organização para a Cooperação Económica e Desenvolvimento
O(G)E (General) State Budget
Orçamento (Geral) do Estado
PAF Performance Assessment Framework
Quadro de Avaliação de Desempenho
PAPs Programme Aid Partners
Parceiros Ajuda aos Programas
PARPA Plan for the Reduction of Absolute Poverty
Plano de Acção para a Redução da Pobreza Absoluta
PARP Plan for the Reduction of Poverty
Plano de Acção para a Redução da Pobreza
PEFA Public Expenditure and Financial Accountability
Despesa Pública e Contabilidade Financeira
PER Public Expenditure Review
Revisão das Despesas Públicas
PESS Sectoral Social and Economic Plan
Plano Económico e Social Sectorial
PES Social and Economic Plan
Plano Económico e Social
PNISA Programa Nacional de Investimento do Sector Agrário
National Agriculture Sector Investment Plan
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | xi
PFM Public Finance Management
Gestão das Finanças Públicas
PRSC Poverty Reduction Support Credit (World Bank)
Crédito para Apoio à Redução da Pobreza (Banco Mundial)
RENAMO Resistência Nacional Moçambicana
Mozambican National Resistance
SAI Supreme Audit Institution
Instituição de Auditoria Suprema
SBS Sector Budget Support
SISTAFE Sistema de Administração Financeira do Estado
State Financial Administration System
SSA Sub-Saharan Africa
SWAP Sector-Wide Approach Programme
Programa de Abordagem dos Sectores
TA Mozambique Supreme Audit Institution / Administrative Court
Tribunal Administrativo
TIA Trabalho de Inquérito Agrícola
Agriculture Survey
TNPA Third National Poverty Assessment
WB World Bank
Banco Mundial
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | xii
Executive Summary
This study has evaluated all of the Budget Support operations undertaken in Mozambique from
2005 to 2012. These operations amount to a resource transfer of US $ 3,354 million - an
annual average disbursement in excess of US $ 400 million, some $16 per annum per head of
the Mozambican population, provided by 19 Development Partners.
In evaluating these operations, our study has addressed three questions:
i. How successful has Budget Support been in providing the means to the
Government of Mozambique to implement its national and sectoral strategies?
ii. How successful has Budget Support been in facilitating improvements in the
efficiency and effectiveness of these national and sectoral strategies?
iii. As a consequence, how successful has Budget Support been in attaining
successful outcomes and impacts on growth and poverty reduction?
Overview of Findings
The primary contribution of Budget support has been in the form of increased funding. In
relation to financial inputs, Budget Support inputs have been important and efficiently
delivered. Disbursements during the 8-year evaluation period were substantial both in fiscal
terms – where they represented on average 15 % of public spending, and as a proportion of
total ODA, where they comprised an average of 30 %. Annual predictability of budget support
has been significantly better than other other aid modalities. The objectives of in-year
predictability and of “front-loading” of disbursements were also largely fulfilled during the
evaluation period.
The inputs provided by the four common funds supporting PFM & Governance functions,
alongside funding lines linked more directly to Budget Support operations, amounted to a
significant TA/ capacity-building input within the Budget Support package. Government
stakeholders were positive in their appreciation of these inputs, both with regard to their
relevance and the relative efficiency with which they were provided. Moreover, there is
evidence that these inputs facilitated the production of the “induced outputs” targeted by
Budget Support, notably with regard to PFM reforms and the strengthening of the capacities of
the Autoridade Tributária de Moçambique (ATM), the Tribunal Administrativo and the Inspeção
Geral de Finanças.
Budget Support is no longer exerting a significant influence on the overall effectiveness of aid.
In the early years of the evaluation period and in the preceding years, the Budget Support
process was instrumental in generating support for aid effectiveness principles. A key aspect of
this was the commitment by the Programme Aid Partners to monitor their own performance
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | xiii
through the PAPs’ Performance Assessment Framework, which was widely heralded as major
step towards mutual accountability and highly valued by the Government of Mozambique.
However, since 2009, performance against these indicators has plummeted. A firm renewal of
the commitment to these objectives and a re-investment by the Government of Mozambique
and its Development Partners in the aid effectiveness principles, will be required for
performance to improve again.
Budget Support has supported the creation of an effective structure for dialogue, based upon
the definition of policy targets and a framework of annual monitoring, comprising sector
reviews and a national level policy dialogue.
The Government Performance Assessment Framework (PAF) fulfils its intended purpose
reasonably effectively and compares well with the PAFs used in other Budget Support recipient
countries. It provides a relatively strategic set of measureable targets, which have worked as a
focus for dialogue. The number of targets is broadly appropriate and the process of assessment
does not entail excessive transaction costs. However, the PAF has inherent limitations, and
requires complementary frameworks in order to work effectively: these complementary
dialogue frameworks are in need of further development.
The main structural weakness of the policy dialogue process lies precisely within the policy
development process, which should be situated at the sector and thematic level. There are
several problems of strategic importance to the PARP agenda – such as low productivity in
agriculture, high levels of malnutrition, a poor enabling environment for doing business - where
effective policy responses have not been generated. GBS has had only a limited influence on
policy development processes at this level.
Through its funding contribution, Budget Support had an important influence on macroeconomic
management. Good macroeconomic management, combined with increased allocations to the
priority sectors, allowed Budget Support to have an important influence on overall growth. GDP
growth remained high throughout the evaluation period, averaging 7.3 % annual growth.
Budget Support funds have enabled the government to increase development spending without
recourse to domestic borrowing, and in this way has supported economic growth. Budget
Support facilitated an expansion in domestically financed development spending from 3.2 % of
GDP in 2005 to 6.1 % in 2012. By avoiding recourse to borrowing, the Government restrained
its demands on the domestic banking sector and facilitated the increase in the ratio of private
sector credit to GDP.
Total spending on the priority sectors designated in PARPA & PARP has more than quadrupled
in nominal Meticais terms over the evaluation period, increasing by slightly more than 7
percentage points of GDP. As a percentage of total expenditure, these priority sectors have
increased their share from 61 % to just over 67 % of total spending.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | xiv
The major contribution of GBS funding has been to support the expansion of public spending in
the education and good governance sectors, alongside a lesser contribution to the agriculture
sector. We draw this conclusion from an analysis of trends in funding sources and sector
expenditure shares, making assumptions that reflect the processes of budgetary negotiation
and decision-making in Mozambique.
By contrast the relative share of government funding (and thus of Budget Support funding) to
the Health sector declined during the period, reflecting a rational response by Government to
the increases in funding from vertical funds and, to a lesser extent, from the PROSAUDE
common basket fund. These big increases in funding have brought important improvements in
health outcomes but have also generated inefficiencies in resource management, whilst also
raising concerns over the sustainability of the delivery structures and the large numbers of
staff (in excess of 5,000 at end 2013) being financed outside of the national budget
framework. Although the health sector has been a component of the GBS dialogue, evidence
does not suggest it had any significant influence on health sector developments.
There has been continued progress made in PFM reform and, to a lesser extent, in governance.
The fact that PFM reform and improvements in governance, especially the fight against
corruption, were systematically discussed and reviewed on an annual basis through the Budget
Support dialogue is likely to have been moderately important to the continued progress made in
these areas.
There is strong evidence that the combination of GBS funding through the budget with targeted
support to institutions of accountability such as the NAO, the Parliament, CSOs and the media
has created more transparency. Targeted support through common funds has been
fundamental to this achievement but Good Governance is one of two sectors that appear as the
primary beneficiaries of Budget Support funds.
Education was designated as a ‘priority sector’ in PARPA/ PARP and has throughout the
evaluation period commanded the highest share of the national budget. This has permitted
dramatic improvements in education provision during the evaluation period:
40 % more children were in primary school in 2012 in comparison with 2004.
The largest gains in the observed increase in primary enrolment were for poorer
households.
17.3% of children of the relevant age group were in secondary education in
2008/09, compared with a rate of 6.1% in 2002/03. Pupil numbers were 65 %
higher in 2012 as compared with 2004.
In terms of provincial trends, progress was made across all provinces both in
primary and secondary enrolment.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | xv
The primary enrolment rate for boys in 2008/9 was only 4% higher than for girls,
indicating a small and gradually reducing gender gap at this level of schooling,
although at secondary level the gender gap is much larger, at 40%.
Funding through Budget Support has played a key role in funding the expansion of the
education sector. The expanded levels of provision could not have been financed in the absence
of General Budget Support. Funding through the FASE common basket funding also contributed,
and FASE processes were instrumantal in setting the policy framework for this expansion.
Our overall conclusion is that Budget Support has been fundamentally successful. It has made
possible a major expansion in education provision, whilst also supporting economic growth and
macroeconomic stability and facilitating steady improvements in the quality of public financial
management and, to a lesser extent, in governance.. These are major achievements, which fully
justify the risks which have been taken in providing Budget Support.
These achievements are still more impressive if one considers the difficult context for Budget
Support. The first evaluation of Budget Support in Mozambique, which covered the period up to
2004, saw this new mechanism operating in a relatively favourable context. The period here
evaluated – 2005-2012, presented a significantly more challenging context:
It has been a period of increasing scepticism over Budget Support from the side of
the Development Partners and increasing disappointment with the aspirations of
the Paris Declaration, with which Budget Support was associated. This has
reflected itself in pressures to obtain quick results, to reduce investment in
institutional development and aid effectiveness work, and to be increasingly risk
averse.
It has been a period of increasing concentration of political power in Mozambique,
which has distanced government from citizens and reduced the need (and perhaps
the willingness) to listen to criticism and to alternative policy voices.
These factors have made more difficult the dialogue between government and the
Budget Support providers by reducing the commitment – on both sides – to the
joint process enshrined in the MoU for Budget Support.
Finally, the public policy agenda has been more technically and politically
demanding.
Future prospects for Budget Support will depend critically on political developments during the
2014 election year but there are reasons to believe that the future context will be more
favourable:
Mozambique’s forthcoming resource boom remains the biggest threat, opening
possibilities for unsustainable foreign borrowing, and creating a potential incentive for
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | xvi
less accountable and less transparent government. Resource extraction will not bring
significant domestic revenues to Mozambique before 2020 but it has already opened
up the opportunity for borrowing against these future earnings. The $850 million
recently borrowed by the Government for tuna fishing vessels for EMATUM and for
naval patrol vessels is an indication of the sort of commercial loans potentially
available. However, at an interest rate of 8.5 %, this is expensive financing and further
borrowing at these levels could generate an unsustainable debt burden, particularly if
the resulting investments did not themselves generate income. In addition, the
investments associated with the resource boom will create significant opportunities for
Mozambican businesses in the construction industry and in service activities. There are
powerful incentives for the political elite to secure preferential access to these
opportunities, and to promote institutional arrangements, which avoid transparency
and public accountability. To date, work on the design of the legal and fiscal
infrastructure for the extraction industries has been undertaken in close partnership
with the IMF and has drawn on a wide range of international advice. Laudable efforts
have been made to reach EITI compliance. It is to be hoped that this approach
continues and that government leaders, officials, CSOs and citizens can work together
with their international partners to establish a mutually beneficial set of arrangements
for managing the resource boom.
Due to the rise of the MDM as a viable third party, the 2014 elections promise to be
more tightly contested than in the past, potentially heralding a period of more
responsive government. Frelimo as the incumbent government and the most
established party will clearly hold the upper hand but it will need to be more mindful of
voters’ concerns than it has been in the past, if it is to avoid significant electoral
losses.
Steady economic growth and fast increases in domestic revenue have created a degree
of fiscal space, which Mozambique has never experienced before. This permits the
Government to contemplate an ambitious programme of public spending targeted on
poverty reduction. Budget Support could allow such a programme to be quickly scaled
up, without serious risk to long-term fiscal sustainability.
The increasing demand for evidence-based aid strategies, combined with the
availability of this evaluation creates a better framework for the design of effective
Budget Support to Mozambique, building upon the favourable fiscal position.
Recommendations
The challenge for Mozambique and its Development Partners is to find ways of achieving more
with Budget Support in the future – in particular to find ways of making a greater impact on
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | xvii
poverty. There has been no statistically significant change in poverty between the early and the
late 2000s, with 54.1% of the population reported below the poverty line in 2003 and 54.7% in
2009, according to the official poverty figures.
With this as the overarching goal, we make a set of recommendations for the joint
consideration of the Mozambican authorities and its Development Partners. These
recommendations need further refinement and development: their implementation will
therefore require a shared dialogue between the Mozambican Authorities and the G-19, in
which analysis and ideas are exchanged and a set of concrete actions is agreed and
implemented. We recommend that this dialogue, and the resulting action plan, should be
clustered around four objectives:
Renewing the commitment to the Budget Support process and the related aid
effectiveness agenda based upon realistic expectations not only of what it may
achieve but also of the investment of time and effort, which is required to make it
work.
Revitalising the Budget Support policy dialogue, establishing a more streamlined
central process for assessing and reviewing performance, and creating a stronger
framework for sectoral policy analysis and dialogue, aimed at stimulating good
policy ideas and actions in the areas where they are currently lacking.
Consolidating and deepening the progress achieved in education, in macroeconomic
management, in PFM reform and in the improvement of transparency and
accountability.
Focussing attention on poverty reduction through targeted programmes to reduce
the incidence of childhood malnutrition, increase the access of small farmers to
fertiliser and other inputs, and to create a conducive climate for business and
employment growth.
Renewing the commitment to the Budget Support process
Within this objective, the basic goal would be to first stop and then reverse the disinvestment
in the Budget Support process, identified by so many stakeholders in Mozambique. A potential
way forward to obtain a renewal of the political commitment to Budget Support by the GoM
and the G-19 members would be a structured campaign at the technical level by the GoM and
the G-19 to present a more realistic message to their political leaders and supporting
constituencies about what Budget Support can and cannot achieve.
An obvious problem in the past has been that the expectations held for Budget Support have
been unrealistically high. What Budget Support can do is firstly to provide funding for areas,
which are clear shared priorities and where tried and tested service delivery strategies exist –
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | xviii
such as expansion of education coverage and quality – and secondly to support and guide those
institutional reforms for which there is a government commitment.
Awareness also needs to be re-established regarding the commitments which Budget Support
entails both for GoM – in terms of transparency, exchange of information, an openness to
dialogue – and for the G-19 – in terms of a commitment to predictable funding, to consistency
in expectations, consistency in the treatment of problems, and to aid effectiveness principles.
The on-going revision of the MoU for Budget Support provides an excellent opportunity to lay
down common rules, regulations and procedures more carefully.
As a part of this process, there will also need to be a commitment – and a related set of
actions – to revitalise the processes of coordination across the G-19. An internal examination
of the reasons for the deterioration in the effectiveness of coordination is recommended, based
on a participatory process – potentially managed by an external facilitator.
Revitalising the Policy Dialogue process
Policy dialogue lies at the heart of Budget Support. It is essential that it should be structured
to be strategic, problem-solving and efficient. At present, the combined structure of central-
level, sectoral and thematic groups within the dialogue structure is not achieving these three
objectives as effectively as it might. Again, we would recommend an internally led,
participatory diagnostic process, focused at two levels:
At the sectoral level, mechanisms need to be found to support effective processes
of research, debate and policy development. These mechanisms need to be
structured so as to strengthen policy development processes in the strategic areas,
where they appear to be weak – such as in the design and implementation of
agricultural policy. A structured, multi-annual programme of research and
evaluation guided through the Budget Support process should probably be created
as part of this mechanism. In addition, some facility could be created for bringing
into problematic sectors external facilitators, who might help to bring a greater
problem-focus into debate. Tanzania’s “Big Fast Results” initiative, led by the
President’s Office provides a good example of such a facility.
At the central level, efforts need to be made to streamline processes and increase
efficiency. Attention should be given to correcting the three areas of inefficiency in
the current arrangements, which relate to (i) the annual review framework created
to mirror the PARP structure, which has inadvertently introduced new transaction
costs and over-complicated the dialogue structures; (ii) the continued lack of
precision in the definition of underlying principles and the methods to be used to
assess and monitor them; and (iii) the size of the PAPs’ PAF and the overly
ambitious targets established in the recent past for these indicators.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | xix
Consolidating and deepening gains
Continued attention must be given to the success areas identified by this evaluation, while
adapting to the new challenges with regard to macroeconomic management, PFM reform,
governance and education.
For macroeconomic management, it will mean greater attention to the framework
for investment selection and management and to the creation of the legal and
institutional framework for managing future revenues from the extractive
industries.
In relation to governance, the challenge is to convert gains in transparency into
tangible gains in accountability.
For PFM reform, the challenge is both to strengthen the design of the PFM Reform
Vision and to reinforce the institutional arrangements for the coordination of
reforms and for the provision of support to reforms.
In relation to education, the challenge is to strengthen the degree of attention
given to education quality issues and to dedicate more attention to poor-
performing schools.
Focussing on targeted programmes to reduce poverty
Finally, we recommend an expanded programme of Budget Support focused on specific
expenditure programmes and policies that might impact on poverty in the short to medium
term. The main shortcoming of Budget Support in the evaluation period has been the failure to
reduce income poverty. However, this is not just a failing of Budget Support but of government
policy as a whole.
The fiscal space now available to Mozambique permits the Government to contemplate an
ambitious programme of public spending targeted on poverty reduction. Expanded provision of
Budget Support could allow such a programme to be quickly scaled up, without serious risk to
long-term fiscal sustainability. In addition to expanded social spending and investment in
physical infrastructure as in the past, we would recommend new programmes of spending and
policy reform focused on three areas:
A programme of social transfers and educational inputs targeted to address
malnutrition.
A programme to improve availability of fertilisers and inputs for small and medium
farmers.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | xx
A programme of policy measures, conducted in close collaboration with the private
sector, aimed at improving the business environment and promoting employment
growth.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 21
1. Introduction
1. This Final Report is submitted on behalf of the ADE Consortium by a team led by ITAD, UK
in association with Fiscus Limited, UK and MB Consulting, Mozambique. It draws on an
extensive documentary review, detailed data analysis – including econometric analysis –
and a wide range of interviews and focus group discussions undertaken during field work in
Maputo, and in Gaza and Zambezia provinces over October - November 2013. The
Evaluation Management Committee, comprising representatives of the Government of
Mozambique (GoM) and of the Programme Aid Partners (PAPs), reviewed the first draft of
this report and circulated it for wider comments from the Reference Group and the full set
of Development Partners (DPs) providing Budget Support to Mozambique. Where
appropriate, the comments received have been addressed by the evaluation team, through
relevant corrections and additions – notably through the addition of further sectoral
analysis for agriculture, education and health and through fuller development of the
recommendations of the evaluation. With these corrections now incorporated, this now
comprises the final version of the Evalution Report.
2. This assignment has formed part of a wider process of evaluation of the effects of budget
support at the international level. It has built upon evaluations of budget support in Mali,
Tunisia, Zambia, Tanzania and South Africa, completed over 2011 - 2013, and an
evaluation of budget support to Morocco, due to be finalised in 2014. In common with
these evaluations, it has applied the OECD-DAC methodology for the evaluation of budget
support.
1.1. Scope and Objectives of the Evaluation
3. The principal objective of the evaluation has been to assess to what extent budget support
in Mozambique has contributed to sustainable results on growth and poverty reduction. The
evaluation has assessed the causal linkages between general and sector budget support
and changes at the outcome and the impact level, considering the effects of the fund flows
but also those of the related policy dialogue processes, and the associated technical
assistance and capacity-building support. The evaluation thus provides answers to three
related questions:
i. How successful have GBS and SBS been in providing the means to the Government
of Mozambique to implement its national and sectoral strategies?
ii. How successful have GBS and SBS been in facilitating improvements in the
efficiency and effectiveness of these national and sectoral strategies?
iii. As a consequence, how successful have GBS and SBS been in attaining successful
outcomes and impacts on growth and poverty reduction?
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 22
4. This has been an independent evaluation, undertaken on behalf of the Government of
Mozambique and the 19 Development Partners who have provided General or Sector
Budget Support over 2005-2012. In particular, the evaluation has covered all General and
Sector Budget Support1 operations over the period 2005-2012, provided by the 19
Development Partners who were signatories of the March 2009 Memorandum of
Understanding on Budget Support – namely, the African Development Bank (AfDB), Austria,
Belgium, Canada, Denmark, the European Commission, Finland, France, Germany, Ireland,
Italy, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, the UK and the World
Bank2.
5. In aggregate, Budget Support disbursements over financial years 2005 to 2012 have
comprised some US $ 3,354 million - that is an annual average disbursement of more than
US $ 400 million, approximately equivalent to US $ 16 per capita. Thus, Budget Support has
been of major importance for Mozambique, comprising an average of 15 % of public
spending throughout the 8-year period, reaching a peak of 18.8 % in 2006. Despite the fact
that, with the growth of domestic revenue and the fall in Budget Support disbursements, its
significance declined to 8.2 % of public spending in 2012, these figures place Mozambique
amongst a select group of 4 countries3 for which Budget Support has represented an
annual average of over 10 % of public spending in the period 2004 -2010.
6. Given the continuing importance of Budget Support, a key objective of the evaluation has
been to provide a forward-looking agenda. It has taken stock of what has been achieved,
whilst also explicitly identifying the shortcomings, in the design and implementation of
Budget Support. The objective has been to draw out lessons for the future: thus, the final
chapters present the conclusions and recommendations which have been developed
relating to:
The conditions under which GBS or SBS are likely to have positive effects in
Mozambique, and the probable intensity and nature of those effects;
1 The definitions of General and Sector Budget Support are provided in OECD-DAC (2006): ‘Budget Support is defined as a method of financing a partner country’s budget through a transfer of resources from an external financing agency to the partner government’s national treasury. The funds thus transferred are managed in accordance with the recipient’s budgetary procedures. […..] In the case of general budget support, the dialogue between donors and partner governments focuses on overall policy and budget priorities, whereas for sector budget support the focus is on sector specific concerns.’
2 Volume Two provides summary details for each of the Budget Support operations, for which detailed documentation was available (Financing Agreements, etc).
3 These are Burkina Faso, Mozambique, Rwanda, and Tanzania. (IOB, 2012)
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 23
The improvements which should be made in the policies and practices of the
Development Partners (DPs), so as to maximise the impact of future Budget
Support operations in Mozambique and elsewhere;
The actions which might be taken by the Government of Mozambique (GoM) in
order to address the constraints arising from its policies, institutional structures
and administrative arrangements, and thus lift the impediments limiting the
effectiveness and impact of the public policy and spending actions financed by
Budget Support and by Mozambican tax-payers, via the national Budget.
1.2. The Evaluation Framework and Process
7. The evaluation has followed the “three-step” methodology, established by the OECD-DAC’s
evaluation network for the evaluation of Budget Support. This methodology is described in
the OECD-DAC’s methodological approach paper,4, and presented in more detail in the
Revised Inception Report for this evaluation (September, 2013).
The Intervention Logic
8. The evaluation methodology proposes an Intervention Logic, in which the outputs,
outcomes and impacts of Budget Support are generated via the policies, budgetary
measures and institutional actions implemented by the recipient government, as a
consequence of Budget Support:
‘…Budget Support is not a development programme per se, but an aid modality that
supports the development strategy of the beneficiary government. …..it produces a number
of direct or immediate outputs, which, under certain conditions, contribute to changes in
government policies and government budgets. The changes thus induced in government
outputs should, in turn, contribute to development outcomes and impacts’.
9. The Intervention Logic is thus based on a set of structured relationships at five levels as
follows:
Level One, which assesses the Inputs of GBS and SBS, including policy dialogue and
capacity-building inputs as well as fund flows.
Level Two, which details the Direct Outputs generated by the interaction of budget
support with other aid modalities.
Level Three, which documents the Induced Outputs produced by Government as a
consequence of the interaction of budget support (in complement with other
modalities) with the national policy, budgetary and service delivery processes.
4 OECD-DAC Evaluation Network (2012), Evaluating Budget Support: Methodological Approach, OECD-
DAC, Paris..
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 24
Level Four, which records the Outcomes of Government policies and spending
actions, in terms of changes in the utilisation of public services by the public and
changes in private sector behaviour (such as investment) due to government
regulatory actions. At this level, government actions are interacting with wider
social and economic forces.
Level Five, which records the wider Impacts of these processes, in terms of societal
and economic processes, notably increased economic growth, reduced income
poverty, and diminished non-income poverty (reduced social exclusion, improved
education, health and social welfare).
The “Three Step” Approach
10. In order to avoid any implicit assumption of causality between Budget Support inputs and
final outcomes, the methodology separates the analysis into three steps, as illustrated in
Figure 1:
Step One covers levels 1-3. It provides the basis for understanding how Budget
Support has been inserted into the public spending and policy making process and
with what effects (direct and induced outputs), given the influence of other aid
modalities and internal government processes working alongside Budget Support.
Step Two begins from an identification of the most significant outcomes and
impacts related to the implementation of the Government’s national and sectoral
development strategies (levels 4 & 5) and then examines – through a combination
of econometric and qualitative analysis – what have been the primary
determinants of those outcomes and impacts.
Step Three brings together the findings from Steps One and Two, identifying which
of the “induced outputs” of Budget Support identified in Step One also feature
amongst the primary determinants of outcomes and impacts identified through
Step Two. In this way, it permits an assessment of the contribution of Budget
Support to final outcomes and impacts and an identification of the key links and
gaps in the Intervention Logic. Hence, Step 3 asks whether Budget Support has
succeeded in inducing the expected outputs and to what extent those induced
outputs have generated positive changes in outcomes and impacts.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 25
Figure 1: The 5 levels and the 3 Steps of the Evaluation methodology
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 26
Figure 2: Budget Support Intervention Logic in Mozambique and proposed Evaluation Questions
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 27
11. In adapting the Intervention Logic (IL) to the Mozambican case, the evaluation team have
considered the relevance and applicability of the range of inputs, direct outputs, induced
outputs, outcomes and impacts identified in the standardised IL. The results of these
considerations are captured in Figure 2, which presents the adapted version of the IL.
Specifically, we have concluded:
That the scale of Budget Support and the range of inputs provided in Mozambique
(funds, policy dialogue, and related support to capacity-building) are consistent
with the inputs foreseen in the standardised framework at Level 1;
That the scale of external assistance in Mozambique justifies the analysis of the
interactions between Budget Support and other modalities foreseen at Level 2;
That the range of issues covered by the PAF and the associated dialogue
framework justifies a focus on the four core induced outputs identified at Level 3.
That detailed attention should be directed, at Levels 4 & 5, to growth and income
poverty (including the influence of the agriculture sector) and to non-income
poverty, as captured in final outcomes and impacts within health and education.
1.3. Evaluation questions
12. The Evaluation framework is broken down into 12 Evaluation Questions (EQs). These are
presented in Box 1 below. Volume Two presents these EQs, together with the proposed
Judgement Criteria for each EQ, as well as the summary answers we have reached and the
sources of evidence for these answers, and an assessment of the robustness of that
evidence. It also includes a four-way classification of the quality of the evidence, explained
in Table 1 below.
Table 1: Criteria for ranking of quality of Evidence for each Evaluation Question
Ranking of Evidence
Explanation of ranking of quality of evidence
Strong The finding is consistently supported by a range of evidence sources, including documentary sources, quantitative analysis and qualitative evidence (i.e. there is very good triangulation); or the evidence sources, while not comprehensive, are of high quality and reliable to draw a conclusion (e.g. strong quantitative evidence with adequate sample sizes and no major data quality or reliability issues; or a wide range of reliable qualitative sources, across which there is good triangulation).
More than satisfactory
There are at least two different sources of evidence with good triangulation, but the coverage of the evidence is not complete.
Indicative but not conclusive
There is only one evidence source of good quality, and no triangulation with other sources of evidence.
Weak There is no triangulation and/ or evidence is limited to a single source.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 28
13. A number of judgements were made in the choice of Evaluation Questions, so as to capture
the specifics of the Mozambican situation, whilst remaining true to the overall
methodology:
The number of Evaluation Questions was restricted to 12 so as to keep within a
manageable magnitude, avoiding dispersion of evaluation efforts.
Following discussions with the Evaluation Management Group, an explicit question
was incorporated at Level One (EQ 1.2) to capture the relevance of the evolving
context for Budget Support, and to consider whether the design changes made to
budget support operations have been appropriate to this evolving context.
With regard to policy dialogue, the Issue Paper (Caputo et al, 2008) proposes two
evaluation questions – one situated at Level 2 relating to the establishment of a
framework for policy dialogue linked to Budget Support and another under Level 3,
relating to overall improvements in policy processes and structures. While
conceptually it is useful to distinguish between the policy structures created for
budget support and the wider structures used by Government, in practice these can
more easily be dealt with as a single EQ, which we have here located at Level 2
(EQ 2.3.).
Reforms to promote governance and democratic accountability have been
essential elements of the policy dialogue. Moreover, ‘the active fight against
corruption’ is monitored as one of the underlying principles for Budget Support
included in both the 2004 and 2009 Memoranda of Understanding. Therefore,
improvements in the quality of governance and accountability are addressed as
‘induced outputs’ in EQ 3.4.
In order to ensure an integrated and holistic analysis, the issues to be examined
under Step Two have been grouped in two EQs, one focused on income poverty
(and including attention to the agriculture sector) and one focused on non-income
poverty (with a primary focus on education outcomes).
Gender issues have been dealt with explicitly within these two outcome/ impact
questions, (EQs 4.1 & 4.2) rather than as a special “cross-cutting theme”. A similar
approach has been adopted in respect of equity issues.
Although all the information necessary for Step Three is generated by the EQs
related to Steps One and Two, an additional evaluation question was added for
Step Three (EQ 5.1) in order to ensure a clear structure for the synthesis process.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 29
Box 1 Evaluation Questions
A. Step One
EQ. 1.1: What was the scale of Budget Support over the period and the mix between GBS & SBS?
What was the mix of inputs provided (funds, TA & capacity-building, policy dialogue) and was it
consistent with those envisaged in the Budget Support agreements, and in the 2004 & 2009
Memoranda of Understanding?
EQ 1.2: How has the context for Budget Support evolved in terms of economic and political
developments, as well as trends in international development policies? Have the changes made in
the scale and in the design and implementation arrangements for Budget Support operations,
including the definition of underlying principles, been relevant to the evolving context?
EQ 2.1: To what extent has Budget Support contributed to increasing the value and the proportion
of external funds managed through the national budget process? How far has this contributed in
turn to increasing the overall predictability of external resource funding for government
activities?
EQ 2.2: To what extent has Budget Support contributed to increased harmonisation of external
aid as a whole and to the reduction of transaction costs per unit of aid provided? Has Budget
Support contributed more or less than other modalities in these respects?
EQ 2.3: To what extent has Budget Support contributed to sustainable improvements in the
processes & methods of policy development and monitoring, in terms of institutionalising
efficient frameworks, which involve relevant stakeholders, focus on strategy and results, and
help to resolve policy dilemmas?
EQ 3.1: To what extent has Budget Support contributed to improvements in the quality of
macroeconomic management and in the effectiveness of domestic revenue mobilisation?
EQ 3.2: To what extent has Budget Support contributed to improvements in the quality of Public
Finance Management (PFM)?
EQ 3.3: To what extent has Budget Support contributed to changes in sector policies and in public
expenditure allocations and with what consequences for the composition of outputs?
EQ 3.4: To what extent has Budget Support contributed to improvements in the quality of
governance and accountability, particularly with regard to the roles of Parliament, Civil Society,
the Tribunal Administrativo and the Anti-Corruption agencies as “watch-dogs” of the Executive.
B. Step Two
EQ 4.1: How has the economy performed in terms of investment and growth and what have been
the effects on income poverty and income distribution? What have been the main determinants of
such changes? In particular, what has been the contribution of the Agriculture sector?
EQ 4.2: How has non-income poverty evolved over the period? In particular, how have the key
outcome and impact indicators evolved in the education and health sectors in aggregate, by
gender and by province? What have been the main determinants of the changes identified?
C. Step Three
EQ 5.1: To what extent have the direct or induced outputs of Budget Support contributed to the
results identified at the outcome and impact levels? To what extent have Budget Support
operations in Mozambique over 2005 to 2012 been effective and have they generated
sustainable impacts?
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 30
1.4. Structure of the Report
14. This introductory chapter has aimed to provide essential background information on the
objectives and methodology of the evaluation. Subsequent chapters follow the structure of
the Intervention Logic underlying the evaluation framework:
Chapter Two examines the context for Budget Support in Mozambique (EQ 1.2). It
considers the economic and political situation at the outset and as it has evolved
over the period of the evaluation, assessing how this may have influenced the
effectiveness of Budget Support. Consideration is also given to the evolution of
Development Partners’ policies on Budget Support and their influence on the
Mozambican case.
Chapter Three addresses levels 1 and 2 of the evaluation framework, asking how
successful Budget Support has been in providing the Government with the
necessary means to implement the national poverty reduction strategy.
Chapter Four completes Step One of the analysis, considering how far Budget
Support has facilitated improvements in Government policies and practices. It thus
considers the “induced outputs” of Budget Support, which comprise level 3 of the
framework.
Chapters Five, Six and Seven report on Step Two of the analysis. They examine the
potential outcomes and impacts of Budget Support, in terms of growth, income
poverty and non-income poverty, analyse how growth and poverty trends relate to
developments in the agriculture sector, and finally consider the relationship of
education sector outcomes to corresponding sector outputs, drawing on a detailed
econometric analysis.
Chapter Eight then reports on Step Three of the analysis, summarising the overall
conclusions on the contribution of Budget Support within the period, and presents
the recommendations of the evaluation team.
The Annexes in Volume Two provide complementary information, including notably
a matrix of responses to the Evaluation Questions and a Summary of the Budget
Support operations evaluated, as well as a statistical appendix, a bibliography, and
a list of persons met.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 31
Figure 3: Map of Mozambique
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 32
2. The Context for Budget Support in Mozambique: help or hindrance?
15. This chapter addresses a specific aspect of Level One of the evaluation framework, ,which
is considered within Evaluation Question 1.2. It examines the nature of the political and
economic context for Budget Support in Mozambique, at the outset and as it has evolved
over the evaluation period.. It considers the key aspects of the Mozambican context and
also the relevant aspects of the international context – notably the changing priorities and
interests of Mozambique’s Development Partners. The key question asked is to what extent
the context may be judged to have been a help or a hindrance to the effective
implementation of Budget Support, considering how that balance may have changed over
the course of the evaluation period.
16. Therefore, in this chapter, we consider four aspects:
The socio-economic context and how it has changed over the evaluation period.
The evolving framework of democratic accountability.
The fluctuations in the policies, priorities and attitudes of the Development
Partners.
Implications for the evaluation and for the future of Budget Support in
Mozambique.
2.1. The evolving socio-economic context of Mozambique
17. From the beginning of the liberation war in 1964, Mozambique experienced nearly 30 years
of violent strife, social and political disruption, and economic crisis. The signing of the
General Peace Accords (GPA) in 1992 heralded a period of almost uninterrupted growth in
excess of 8 % per annum (World Bank & IMF, 2005). At the same time, the point of
departure was not a favourable one: in the early 1990s, Mozambique faced among the
deepest problems of absolute poverty, in a largely subsistence economy, with a long
history of bitter civil war, during which the rudimentary public services which had existed
had been eroded and debt had mounted. (Batley et al, 2006).
18. Supported by high volumes of foreign aid, debt forgiveness under HIPC and new
investments in mega-projects (notably the Mozal aluminium smelting plant), Mozambique
managed a remarkable recovery. The combination of the return to peace, the re-
establishment of government services and fast economic growth led to an impressive
reduction in the level of absolute poverty, from 69% of the population in 1996-97 to 54%
in 2002-03, thus bringing almost 3 million people above the poverty line. This achievement,
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 33
together with the signing of the Memorandum of Understanding for Budget Support in
2004 and other positive developments in policy making - notably the publication of the
2001-2005 poverty reduction strategy, PARPA I, created a highly favourable context for
Budget Support. In particular, the government enjoyed, at the time, a close and supportive
relationship with its Development Partners, and an apparently positive outlook, not only for
growth but also for poverty reduction.
19. A first evaluation of General Budget Support to Mozambique was undertaken during 2005
and published in May 2006, utilising an early version of the OECD-DAC methodology5. Its
conclusions were, indeed, positive:
‘Our broad conclusion is that this has been a very successful case of donor-government
collaboration, and that GBS has contributed positively to conditions for economic growth
and poverty reduction.’
20. Nevertheless, this first evaluation did stress that, in 2004, there persisted high rates of
poverty, poor health indicators and high rates of illiteracy, and that ‘indicators of access to
education and health services show a poor situation even by comparison with other less
developed countries. The main reason for declining life expectancy is the onset of HIV/ AIDS:
Mozambique now has the eighth highest prevalence rate in the world.’ (Ibid, p.7.) In short, a
more careful interpretation of socio-economic trends in 2004-05 might have highlighted
that 12 years of fast growth in the post-war recovery period had still not been enough to
address in a significant way the indicators of social deprivation prevailing at the time of
the Peace Accords. Moreover, the continuing deficit in physical and human capital
investment, combined with the new threat of HIV/ AIDS, did not bode well for a
continuation of the remarkable reduction in income poverty achieved over 1996-97 to
2002-03, particularly given the challenge of reaching the predominantly rural poor within a
country as vast as Mozambique (approximately 800,000 square kilometres), with an
average population density of only 29 inhabitants per square kilometre (2012).
Evolution of growth and poverty trends during the evaluation period
21. Mozambique has experienced high and sustained growth during the evaluation period at an
average rate of 7.3%. This is illustrated in Figure 4 below that shows that Mozambique’s
economic growth rates have compared favourably with other Sub Saharan African (SSA)
countries, including those ‘frontier economies’ within SSA.
Figure 4: Real GDP for Mozambique in comparison with other ‘frontier economies’
5 Batley, R., Bjornestad, L. & A. Cumbi (2006), Joint evaluation of General Budget Support, 1994 – 2004:
Mozambique Country Report, IDD: Birmingham.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 34
Source: IMF Country Report No. 12/148, June 2012
22. Despite the fast growth achieved, Mozambique’s progress in poverty reduction has been
less impressive. Although two different methodologies have been applied in measuring and
analysing poverty (the official poverty figures, presented in the Third National Poverty
Assessment and by the World Bank in Alfani et al 2012), both tell a consistent story at the
national level: there has been no statistically significant change in poverty between the
early and the late 2000s.
Table 2: Economic & Social Indicators for Mozambique and Tanzania, 2005 and 2012
Mozambique Tanzania
Indicator 2005* 2012* 2005* 2012*
GNI per Capita, Atlas Method (current USA $) a 290 510 380 570
Population, million a 21,01 25,20 38,8 47,7
% Poverty headcount ratio at national poverty line (% of population) a
54.1 (2003)
54.7 (2009)
35.7 (2001)
33.41
(2007)
Life Expectancy at Birth b 48.1 50.7 53.8 60.1
HDI (Score) b 0.287 0.327 0.420 0.446
HDI (Ranking) b 168 185 159 152
Literacy Rate a 48.16 (2003)
50.58 (2009)
69.4 (2002)
67.8 (2010)
Prevalence of undernourishment (% of population) a 40.3 39.2 (2011)
35.1 38.8 (2011)
Depth of the food deficit (kilocalories per person per day) a 310 292 248 274
Prevalence of HIV, total (% of population ages 15-49) a 11.3 11.1 6.3 5.1 * Where 2005 or 2012 data were not available, the most proximate year has been used.
Sources a World Development Indicators 2013; b Human Development Report 2013.
23. Growth and poverty trends and their probable determinants have been analysed as part of
Step Two of the evaluation and our results are presented in Chapter 5. However, an
overview of the trends in socio-economic indicators is provided by a reading of Table 2
above, which presents summary statistics for 2005 and 2012 for key economic and social
indicators. We have provided comparative data for Tanzania, which has quite a number of
REPUBLIC OF MOZAMBIQUE
4 INTERNATIONAL MONETARY FUND
I. BACKGROUND AND RECENT DEVELOPMENTS
1. The Mozambican economy recorded
a strong performance in 2011, showing
little sign of being affected by the global
turmoil. Preliminary data indicate that GDP
rose by about 7 percent, comparing favorably
with the growth rates of Mozambique’s peers.
Growth was broad-based, with agriculture,
mining, transportation and communication,
and financial services registering the fastest
expansion. Activity was also supported by
strong exports and robust investment demand
from megaprojects in the natural resource
sector and the public sector.
2. The authorities’ disinflation policies
yielded commendable results. Twelve-month
inflation decelerated sharply to 2¾ percent in
March 2012, compared to 16½ percent at end-
2010. Core inflation, once hovering in the
double-digit range, fell to 2⅓ percent in
March. This reflects a determined monetary
tightening during 2010-11, which decelerated
broad money growth by two-thirds in 2011
from a year earlier and contributed to the
appreciation of the metical relative to main
partners’ currencies. An appropriately
restrictive fiscal policy stance, especially in the
first half of 2011, was also conducive to
disinflation, as were favorable trends in
international food and energy prices.
3. The external sector performed well,
and Mozambique’s net international
reserves (NIR) were further buttressed.
Preliminary data for 2011 suggest that both
megaprojects’ and traditional exports were
buoyant, although strong import demand for
investment goods resulted in a slight
worsening of the current account deficit, to
13 percent of GDP. Led by megaprojects, FDI
more than doubled relative to 2010, which
supported an NIR accumulation of
US$331 million and a comfortable reserve
coverage (5.2 months of current-year imports).
Mozambique
Sub-Saharan
Africa
Frontier
economies 1
0
2
4
6
8
10
12
14
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Real GDP(Percent change from previous year)
Proj.
.1Frontier economies: Angola, Ghana, Kenya, Mauritius, Mozambique, Senegal, Tanzania, Uganda, Zambia and Zimbabwe.
Mozambique
Frontier
economies
Sub-Saharan
Africa0
5
10
15
20
25
30
35
40
45
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
CPI(End of period, percent change from previous year)
Proj.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 35
similarities with Mozambique, notably in terms of its natural resource endowments, per
capita incomes, fast recent growth rates, and the predominantly rural structure of the
population. However, the two major differences relate to population density (28.7 per
square kilometre in Mozambique vs. 47.5 in Tanzania) and, more importantly, to the record
of 50 years of peace since Independence in Tanzania, as compared with Mozambique’s
legacy of civil war.
24. The comparison with Tanzania is, to put it simply, not flattering for Mozambique. Across
every social indicator, performance in 2005 is worse in Mozambique and remains worse in
2012. Only in relation to literacy and prevalence of under-nourishment6 has the
performance of Mozambique been better than Tanzania within the evaluation period.
25. How are we to interpret this? Clearly, there are likely to have been weaknesses in the
strategic allocation of resources and in the design and implementation of policy in
Mozambique. This is a question we examine in more detail in Chapter 4, considering in
particular whether there is evidence of Budget Support having positive or negative effects
on these processes. Nevertheless, Tanzania is not a particularly good performer amongst
its SSA peers in the design and implementation of policy and in the efficiency of its public
administration systems7. Fundamentally, the explanation is that, in any low income
country, the legacy of civil war leaves a long foot-print in terms of deficiencies in human
capital, in physical infrastructure and in the administrative capacity to deliver services8.
Tanzania is able to perform better than Mozambique across most socio-economic
indicators because of a “delivery infrastructure”, which has been built up over many years.
The “catch-up challenge” for Mozambique was daunting at the outset of the evaluation
period – still more so because of the high prevalence of HIV/ AIDS - and remains daunting
now. By implication, the expectations of the likely impact of public policy actions – and
therefore of the likely impact of Budget Support – need to be tempered to this reality.
Resilience of the economy and future prospects
26. Mozambique has been reasonably unaffected by external economic events over the
evaluation period. The impact of the global financial crisis on Mozambique was relatively
6 World Development Indicators 2013 reports a second indicator, measuring the % of children under 5
determined as malnourished in terms of their weight for age. Data for Mozambique are reported as 18.3 % in 2008 and 11.6% in 2011. If correct, these data suggest a dramatic improvement (for example for the same indicator, prevalence in Tanzania fell from 16.7% in 2004 to 16.2% in 2010), but these data are not consistent with the other indicators for malnourishment shown in Table 2 and show an unusually fast rate of change for such an indicator, suggesting some methodological inconsistency or some inaccuracy.
7 A recent evaluation of Budget Support to Tanzania over 2006 -2012 identified several examples of weaknesses in the design and implementation of public policies. (See Lawson et al, 2013).
8 For example, see Collier (2007), The Bottom Billion, on the long-term costs of civil war in low income countries.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 36
limited, with a fall in economic growth during 2008-2009, as a result of a decline in
foreign direct investment and a collapse in the world commodity markets. This led to
export receipts falling and a decline in private sector financing. However, Mozambique’s
rate of economic growth bounced back reasonably quickly, due to a countercyclical policy
stance, increased donor support, particularly though the IMF’s Exogenous Shocks Facility
(ESF) and an increase in lending by the domestic banking system which compensated for
falls in external financing. By 2010, export levels had recovered and growth rates had
returned to previous levels.
27. Mozambique also experienced economic shocks from increases in international food prices
during 2007 and 2008, when petrol and food prices rose 27 and15 per cent respectively.
This led to large domestic fuel price increases in February 2008, which triggered riots.
However, the economic impact of these events was again only temporary and levels of
inflation fell again, while economic growth rebounded quickly. Similarly, economic growth
was affected by bad weather in late 2007, which resulted in extensive flooding. Despite
this, there was only a short-term impact on economic growth.
28. The stability of growth has therefore been a significant feature of the evaluation period,
especially by comparison with the much more volatile rates of growth experienced over
1995-2005. The structure of growth has also been reasonably stable with agriculture and
services & trade accounting for at least two thirds of GDP throughout the period. (In 2011,
agriculture accounted for 39% of GDP and services & trade for 27%.)
29. The main economic transformation in Mozambique has been the discovery over 2010 –
2012 of very large deposits of exploitable gas in the Rovuma Basin, off the coast of Cabo
Delgado province. Anadarko Petroleum Corporation and the Italian company, ENI, who are
leading the exploration work in this area have stated that there are known reserves in the
Rovuma Basin alone in excess of 100 trillion cubic feet9, which would justify the
construction of a Liquefied Natural Gas (LNG) facility on the coast. Although it has long
been known that Mozambique has rich and varied resources of minerals (titanium, gold,
heavy sands) and energy (coal, gas, hydropower, and probably oil), and exploitation of
some of these had already begun [hydropower from Cahora Bassa, coal in Tete, gas
exported to South Africa by pipeline from the Pande and Temane fields in the Mozambique
Basin (Inhambane/ Sofala)], it is these new discoveries that are considered a real “game-
changer”.
30. The planned investments in the Rovuma basin, combined with the requirements for the
increased exploitation of the gas reserves in the Mozambique Basin, and of coal and heavy
9 The Empresa Nacional de Hidrocarbonetos (ENH) have stated that Mozambique as a whole may have
as much as 250 trillion cubic feet (t.c.f) of reserves. These figures compare with the known reserves of 40 t.c.f and estimated total reserves of 100 t.c.f in Tanzania.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 37
sands amount to total Foreign Direct Investment (FDI) of some US $88 billion over the next
ten years10, more than six times the current GDP. Clearly, this “mining boom” has the
potential to raise per capita income levels dramatically and to provide a major new source
of domestic revenue. Already in 2013, the Mozambican government received significant
revenues from the taxation of capital gains on the re-sale of exploration contracts within
the Rovuma Basin (equivalent to 4.2 % of GDP).
31. However, these are very preliminary estimates, subject to a high margin of uncertainty, in
particular in relation to the timing of investments and the start of the exploitation process.
At present, the earliest estimate for the start of gas extraction from the Rovuma Basin is
2018 but, given the complexity of the infrastructure works which will need to be completed
before extraction may begin (railway lines, ports, LNG installations), this estimate seems
optimistic. Moreover, the flow of domestic revenues resulting from extraction will arrive
with a significant lag. This is because these revenue flows will derive primarily from
production royalties, which, are payable only after the beginning of production, and from
dividend and profit shares due to the Government’s shareholdings in these developments,
which will be depleted by amortisation of investment costs. Moreover, these shareholdings
will themselves need to be paid for from the production royalties, so positive net flows to
Government will only kick in some time after production has commenced.
32. Therefore, significant domestic revenue flows from resource exploitation are unlikely to
occur before 2020 and the Bank of Mozambique, when interviewed, stated that external
funding is likely to be needed at least until then. Nevertheless, the move away from aid-
dependence has been lauded by the President himself11, as well as other commentators, as
one of the important expected benefits of the “mining boom” and any suggestion in the
short-term of a perceived increase in aid dependence through the application of tighter
conditionality will certainly be resisted. This has important implications for the design of
Budget Support and of delivery strategies for aid as a whole.
33. Within the evaluation period, developments in resource extraction have not been of
significant importance either as a source of domestic revenue or as a source of economic
growth, where the sector still only accounted for less than 2% of GDP in 2011 (INE).
However, the design of the future fiscal and institutional framework for natural resource
exploitation has been an aspect of the dialogue within the Budget Support arena and more
widely. In October 2012, after 3 years as a candidate country, Mozambique was granted
the status of a fully compliant member of the Extractive Industries Transparency Initiative
10 This estimate is based on the announced plans of Anadarko, ENI, SASOL and Statol in Gas, Kenmare in
heavy sands and Vale and Rio Tinto in coal. (Internal document of the EU Delegation, June 2012).
11 As reported in Gqada, I. (August, 2013), A boom for whom? Mozambique’s natural gas and the new development opportunity, South African Institute of International Affairs Occasional Paper No. 151; Capetown.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 38
(EITI). This was widely seen as an important step towards the establishment of transparent
contractual and taxation arrangements. Nevertheless, many aspects of the fiscal and
institutional framework remain to be decided.
2.2. The evolving framework of democratic accountability
34. The quality of underlying processes of democratic accountability is an essential aspect of
the context for Budget Support. Where the ‘core’ level of democratic accountability is high
from the outset, then the modest changes which Budget Support might be able to induce
are more likely to have a significant and lasting effect; where this is low, then Budget
Support processes are less likely to be influential. 12
35. What can we say about the nature of democratic accountability within the evaluation
period? We examine three potential indicators of the vitality of the systems of democratic
accountability:
The fairness and competitiveness of the electoral system;
The degree of popular support for the ruling party or coalition;
The existence of a civil society with an active voice.
Fairness and competitiveness in the electoral system
36. Mozambique promulgated a new democratic constitution in November 1990, and following
the peace accords of 1992 has celebrated general elections for the Presidency and the
parliament every five years since the first multi-party election in 1994. Frelimo have been
the consistent winners of these elections, initially under Joaquim Chissano, who was
president for18 years (including 10 under a democratic mandate) and since 2004, under
the presidency of Armando Guebuza who won the 2004 and 2009 elections. Having
completed his statutory two terms, President Guebuza has announced that he will step
down, and in February the Frelimo Central Committee voted for Filipe Nyussi, the current
Defence Minister, as their candidate for the October 2014 presidential elections. .
Table 3: Parliamentary Election results 1994 – 2009 (% of valid votes cast)
1994 1999 2004 2009
Seats Votes % Seats Votes % Seats Votes % Seats Votes %
Mozambique Liberation 129 44.3% 133 48.5% 160 62.0% 191 74.6%
12 We acknowledge the relevance of the debate about the apparently limited role of democratic
accountability within the “Developmental State”, as discussed by Mustaq Khan and Ha-Jong Chang, amongst others. (See also Fritz and Menocal, 2006 for an excellent summary of this literature.) However, within the constitutional framework of Mozambique, it is the institutions of democratic accountability which are intended to provide the checks and balances to control the abuse of power and to guide public decision-making towards efficient and effective solutions. Our objective is simply to provide an indication of the vitality of the systems of democratic accountability during the evaluation period, while recognizing that this analysis is open to a range of interpretations with regard to the implications for politics, and for economic and social policy.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 39
Front (FRELIMO)
Mozambican National Resistance (RENAMO)
112 37.8% 117 38.8% 90 29.7% 51 17.6%
Third most voted party
9 (UD)
5.2% 0 - 0 - 8
(MDM) 3.9%
Total 250 87.3% 250 87.3% 250 91.7% 250 96.1%
Source: AIM – Mozambique News Agency
37. As may be seen from Table 3, in the 1994 and 1999 elections, Renamo achieved a
relatively high percentage of the votes in the parliamentary elections, and still higher
percentages of the presidential vote13, where in 1999, the Renamo candidate, Afonso
Dhlakama, obtained 47.7% of the valid votes cast to Chissano’s 52.3%. The strong
performance of Renamo in each of these elections came as a surprise to many observers
but suggested clearly that there were significant groups in society, who did not support
Frelimo policies. Some commentators observed that right through the transition from a
one-party to a multiparty political system, the Frelimo state had always remained
‘suspended above society’ (Saul, 1993.) However, in subsequent elections, led by Armando
Guebuza, Frelimo were able to win a dominant share of the votes cast, and thus of the
Parliamentary seats, which are decided by proportional representation.
38. Renamo criticised the results of each of these elections for not having been conducted in a
fair and transparent manner by the National Electoral Commission (CNE). In respect of the
1994, 1999 and 2004 elections, observers from the Carter Centre and from the European
Union Election Observation Missions identified several cases of electoral fraud but they
concluded that the electoral shortcomings would not have affected the final result in the
presidential election, and would only have had a minor impact on the distribution of
parliamentary seats.
39. In the 2009 elections, the two opposition parties – Renamo and MDM (Movimento
Democrático de Moçambique) alleged fraud. The EU and other observer groups reported
that voting was conducted in a peaceful and orderly manner but they were highly critical of
many pre-election day processes. In particular, it was pointed out that the CNE’s
disqualification, for ostensibly technical reasons, of MDM candidates’ nomination papers in
9 of the 13 parliamentary constituencies had substantially restricted voter choice14.
Observers also documented election day irregularities, including ballot stuffing and
tabulation fraud at some polling stations, although these distortions were considered
insufficient to have affected the overall result of the election. (Freedom House, 2010).
13 Apart from the 1999 elections, the results for the parliamentary and presidential votes have been
very similar. See AIM - Mozambique News Agency or International IDEA.
14 As a rough indication of the impact this might have had, if MDM had won the same percentage of parliamentary as presidential votes (8.59 %), they would have won some 15-20 seats, as opposed to the 8, they actually won. Whether this would have been at the expense of Frelimo or Renamo is difficult to judge.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 40
40. The reported electoral irregularities, combined with on going concerns about slow progress
in the fight against corruption, prompted the G-19 Budget Support group to report to
Government that it felt that a breach of the underlying principles for Budget Support might
have occurred. This led to the suspension of Budget Support disbursements by 4 DPs and
an intense process of dialogue between Government and the G-19, which became known as
the period of “crispação” (“political tension”). As a result, Government agreed to implement
a Governance Action Plan, whose key elements included the preparation of revisions to the
electoral law, certain measures to promote political inclusivity, and a package of Anti-
Corruption legislation.
41. Notwithstanding the improvements to electoral processes, which may result from the
Governance Action Plan, it is clear that the period of the evaluation has been characterised
by a sharp reduction in the level of political competition. With the 2009 elections, Frelimo
was able to assume a position of real dominance over political institutions. Within a
presidential system, where many judicial and other appointments are made by the
President, this has led to a major centralisation of power. The hierarchical structure of
Frelimo and the strong internal party discipline have reinforced this centralisation.
The degree of popular support for the ruling party
42. Over successive elections, Frelimo has been able to assume a position of political
dominance. This may have been helped in a small way by electoral irregularities. Yet, it is
likely that it would have happened anyway, simply by virtue of the better organisation of
Frelimo in comparison with the opposition, and by virtue of the advantages bestowed by
being the incumbent party in power. However to what extent does it reflect an increasing
level of popular support for Frelimo?
43. In this respect, the data on voter turn-out and numbers of invalid votes over the course of
the four general elections are revealing. These are presented in Table 4 and Figure 5. They
show a sharp decline in voter turn-out from 87.9 % of registered voters in 1994 to only
36.3 % in 2004, rising modestly to 44.4 % in 2009. Simultaneously, there was first a
decline in the number of invalid votes to 8.3 % in 2004 from the 1994 high of 11.7 %,
probably attributable to the learning process following the nation’s first multiparty
election. However, this was then followed by a sharp rise in numbers of invalid votes in the
2009 election to 11.25 %. Casting invalid votes is a classic form of voter protest, and the
2009 outcomes could be interpreted in such a way. High numbers of invalid votes is also
frequently associated with electoral fraud (through the invalidation of votes for opposition
candidates). Certainly, the proportion of invalid votes is higher and, in most years, the voter
turn-out lower than other countries within the region, which are also dominated by large
majority parties.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 41
Table 4: Voter turn-out and invalid votes cast in Mozambique & neighbouring countries
Figure 5: Voter turn-out and invalid votes in Mozambique & neighbouring countries
44. Clearly, these data are open to a range of different interpretations. However, the fact is
that in 2009, Frelimo received the votes of only 2.9 million adults, 29.5 % of the registered
voters, as compared with the 34.4 % who had voted for them in 1994 (2.1 million). In
short, Frelimo now has more political power but less popular support, which opens up
questions about its legitimacy and the degree of respect in which it is held by Mozambican
citizens.
45. It is possible that these data simply reflect dissatisfaction with the political process and
with the political alternatives on offer, rather than dissatisfaction with Frelimo as such.
However, there are other indicators and developments, which suggest a degree of citizen
dissatisfaction with the governing party and perhaps a sense of disenfranchisement:
The violent popular protests that followed the rise in fuel prices and public
transport fees in February 2008 and the combination of food and fuel price rises in
September 2010 suggest firstly a willingness to engage in popular protest which
has not been seen in the past. Secondly, the failure to anticipate this reaction in
advance suggests an increasing distance between the leadership and the people: in
2008 a 50 % increase in public transport fees was imposed, and the ensuing riots
left 4 people dead and more than 100 injured. (Freedom House, 2010.)
The popular perception of corruption has remained consistently high as reported by
Transparency International (See Table 5). More significantly, it is the Police, who
are identified by Mozambicans as the most corrupt public institution. (Transparency
International, 2011).
Parliamentary Elections' Results
1994 87.89%
1999 68.09%
2004 36.34%
2009 44.44%
Source:TheInternationalInstituteforDemocracyandElectoralAssistance(InternationalIDEA)
Mozambique
Year of
elect ion
Voter
Turn- out
Parliamentary Elections' Results
11.70% 1996 59.30% 1995 76.51% 5.50% 1994 86.87% 0.98%
9.60% 2001 70.31% 2.70% 2000 72.77% 4.60% 1999 89.28% 1.50%
8.30% 2006 68.00% 3% 2005 72.52% 5% 2004 76.73% 2%
11.25% 2011 59.29% 0.0405 2010 39.49% 2009 77.30% 1.3%
Source:TheInternationalInstituteforDemocracyandElectoralAssistance(InternationalIDEA)
Year of
elect ion
Voter
Turn- out
I nvalid
votes
Mozambique Uganda Tanzania SouthAfrica
Year of
elect ion
Voter
Turn- out
I nvalid
votes
Year of
elect ion
Voter
Turn- out
I nvalid
votes
I nvalid
votes
0%
2%
4%
6%
8%
10%
12%
14%
('94-'96)
('99-'01)
('04-'06)
('09-'11)
Invalidvotes
Mozambique
Uganda
Tanzania
SouthAfrica
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 42
The report of the Africa Peer Review Mechanism (APRM 2010) points out problems
of social and economic inequalities, and very limited access to resources and
economic opportunities by the majority of the population.
The report on the State of the Union prepared by CESC in 2013 indicates problems
of political intolerance and a strong aversion to criticism on matters of governance
by the government15.
Finally, several important personalities from the political, academic and religious
arena have publicly voiced their concerns about the state of affairs in the country,
particularly regarding governance and the model of wealth distribution in
Mozambique16.
Table 5: Indicators of Political Openness and Perceptions of Corruption, 2005 & 2012
Mozambique Tanzania
Indicator 2005* 2012* 2005* 2012* Political Openness 2005 & 2012 a (indication of the general state of freedom in a country)
Political Rights (1=high freedom; 7=low) 1
3 4 4 3
Civil Liberties (1=high freedom; 7=low) 2
4 3 3 3
Corruption Perceptions Index Score b (0 = highly corrupt; 10 = no corruption)3
2.8 3.1 2.9 3.5
Notes: 1 The ratings process is based on a checklist of 10 political rights questions. Scores are awarded to each of these questions from
which a rating of 1 to 7 is derived, with 1 representing the highest and 7 the lowest level of freedom. 2 The ratings process is based on a checklist of 15 civil rights questions. Again, a rating of 1 to 7 is derived, with 1 representing
the highest and 7 the lowest level of freedom. 3 The CPI measures the degree to which public sector corruption is perceived to exist within a country on a scale of 0 (highly
corrupt) to 10 (very clean).
Sources:
a Freedom In the World 2012, Freedom House; b Corruption Perceptions Index 2012, Transparency International.
The existence of a civil society with an active voice
46. The fact that there are commentators from civil society able and willing to stand up and
voice their concerns over the state of governance is a positive sign. The importance of
these groups has been noted by a number of researchers. Shikhani (2012) states that
‘Mozambique’s socio-cultural system is characterised by the existence of many pressure
groups. They constitute a growing critical mass and play a fundamental role in helping to
deepen democracy in the country.’ Freedom House (2010) reports that ‘religious freedoms
are well respected, and academic freedoms are generally upheld’. It also notes that ‘local
journalists and non-governmental organisations such as the Centre for Public Integrity
15 CESC (2013) gives examples of destruction of flags of opposition parties in the provinces of Manica
and Gaza, the closure of three community radios with programmes focused on monitoring governance, a coordinator of the civil society platform of Barue threatened by the district administrator with expulsion from the district.
16 We refer in particular to Prof. Lourenço do Rosário, Rector of Apolitécnica; Dinis Sengulane, Bishop of Lebombo; Francisco Chimoio, Archbishop of the Catholic Church in Maputo; and Prof. Carlos Nuno Castel Branco of the Institute of Social and Economic Studies (IESE).
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 43
have played a crucial monitoring role by investigating and exposing high-profile corruption
cases.’
47. Since the introduction of multiparty democracy in 1994, new independent media sources
have proliferated. These include several weeklies and the daily, O Pais, a number of
independent and community radio stations, and, more recently, news websites. (Freedom
House, 2010) There is also political debate on issues like the national and municipal
elections on social networking sites such as Facebook. Shikhani (2012) also reports that
‘mobile phones were crucial in the preparations for the events of 5 February 2008 and 1-2
September 2010, with the youth who participated in these events being mobilised by SMS
days before’. In short, there is an active process of circulation of information, ideas and
analysis and significant capacity for social mobilisation.
Table 6: Worldwide Governance Indicators for Voice & Accountability, 2006 & 2012 Country Year Score
(-2.5 to + 2.5) Percentile Ranking
(0 to 100)
Standard Error (for scores)
Lesotho 2006 0.15 52.9 % 0.17
2012 0.04 51.7 % 0.14
Mozambique 2006 -0.08 45.7 % 0.11
2012 -0.18 43.1 % 0.12
Tanzania 2006 -0.20 40.4 % 0.11
2012 -0.22 41.7 % 0.11
Malawi 2006 -0.26 38.9 % 0.12
2012 -0.23 41.2 % 0.11 Source: Worldwide Governance Indicators, 2013
48. As a reflection of these developments, Mozambique scores relatively well on the Worldwide
Governance indicators for Voice and Accountability. As shown in Table 6, its percentile
ranking in 2012 was 43.1 %, which compared well with similar neighbouring countries, as
well as with the average ranking for low income countries (22.5%) and for Sub-Saharan
Africa (31.6 %). There has been a minimal decline in Mozambique’s score against this
indicator between 2006 and 2012 but this falls within the bounds of the standard error. As
reflected in Table 5 above, Freedom House show a modest improvement (from 4 to 3)
between 2005 and 2012 in the ranking for civil liberties, which measures a similar set of
issues.
49. Notwithstanding the Freedom House and WGI rankings, there continue to be shortcomings
in the quality of voice and accountability. While press freedoms are legally protected,
journalists are sometimes harassed or threatened and often practise self-censorship
(Freedom House, 2010). The campaign periods leading up to the 2009 national elections
and the 2013 municipal elections were both marred by a significant level of partisan
violence. Some commentators (CESC, 2013) suggest that the level of harassment and
intimidation of civil society has increased in recent years. This view is supported by a
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 44
number of survey-based indices and, overall, we would conclude that the quality of voice
and accountability during the evaluation period has probably declined. As an illustration of
this, we note the following:
In the 2012 Global Index of Freedom of Expression Mozambique dropped seven
places in the ranking, from 66th in 2005 to 73rd in 2012.
Afrobarometer showed that the number of people who think that they are
completely free to say what they think fell to 41% in 2012, compared to 47% in
200817.
The World Press Freedom Index ranked Mozambique as 73 out of 180 countries in
2013 a considerable fall from a ranking of 49 out of 167 in 2005, with its overall
score falling from 10.5 out of 100 in 2005 to 28.0 in 2013, where zero indicates
the best score and 100 the worst18.
2.3. Fluctuations in the policies, priorities and attitudes of the Development Partners
50. The relative shift in European opinion at public and official levels regarding the desirability
of Budget Support has also represented an important change in the context for Budget
Support during the evaluation period. This is evidenced in the international withdrawal from
budget support operations by the Netherlands, and also in the explicit changes in policies
introduced by the European Commission, Sida, the State Secretariat for Economic Affairs of
Switzerland (SECO) and the Department for International Development (DFID)19. These
changes in official policies have promoted more rigorous assessments of the fiduciary risks
of budget support and attempts to develop more explicit links between disbursement
modalities and the results of Budget Support.
51. Neither the requirement for more rigorous fiduciary risk assessment nor the desire for
more results-based disbursement methodologies need have been problematic for the
implementation of Budget Support in Mozambique. Indeed, greater thoroughness in the
fiduciary risk assessment process and in the monitoring and communication of results
should lead to a better informed dialogue around Budget Support. However, in interviews,
several senior Development Agency staff attributed these policy changes to the desire of
domestic politicians to see more immediate and more tangible results. It was reported that,
as a consequence, these policy changes had often been accompanied by a declining interest
17 http://www.afrobarometer.org/data/data-by-country/mozambique
18 https://rsf.org/index2014/en-africa.php
19 New policy statements on Budget Support have been produced by each of these agencies since 2008. The broader shift in the fashion for Budget Support is documented in IOB (2012) and Tavakoli & Hedger (2011).
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 45
in longer term institutional development processes (including those linked to the aid
effectiveness agenda), by a greater risk aversion, and by increasingly “knee jerk” reactions
to unfavourable events, such as corruption scandals. The evaluation of Budget Support to
Tanzania (Lawson et al, 2013) identified increased evidence of these phenomena within the
Tanzanian Budget Support process in the post 2007 period. Our analysis in Chapter 3
examines whether these phenomena were present in the Mozambican case.
52. Finally, it is worth noting that the global financial crisis has put pressure on public finances
in many OECD countries, and that in some cases this has had a knock-on effect on aid
budgets. The OECD DAC reported in April 2012 that there had been a 3% drop in total ODA
disbursements between 2010 and 2011. As the aid modality which is easiest to “turn off”,
Budget Support may have been disproportionately affected by this. Four of the G-19
members providing Budget Support to Mozambique – Italy, Ireland, Portugal and Spain –
have formally announced cuts in their aid budgets. Two of these – Spain and Portugal –
have temporarily suspended payments of Budget Support, while Ireland and Italy have both
reduced the value of their disbursements. On the other hand, these are relatively small
Budget Support contributors and the three largest contributors to Budget Support in
Mozambique – the World Bank, the EU and DFID – have expanding aid envelopes (through
IDA and EDF) or explicit political commitments to protect aid diisbursements - in the case
of the UK through a commitment to achieving a target of 0.7 % GDP for total ODA. In
short, within the evaluation period, the more relevant factors would appear to be the shift
in Budget Support policies and in political attitudes towards the use of Budget Support. .
2.4. Implications for the evaluation and for future Budget Support
53. Any evaluation of an aid intervention mechanism must consider both the quality and
relative efficiency of the mechanism itself and the nature of its interaction with the
operating context. The outcomes of Budget Support are the consequence of the
combination of the mechanism and the context20.
Implications for the evaluation
54. The first evaluation of Budget Support in Mozambique (Batley et el, 2006), which covered
the period up to 2004, saw this new mechanism operating in a relatively favourable
context. By comparison, the period here evaluated – 2005-2012, presented a significantly
more challenging context in, at least, four respects21:
20 See Pawson and Tilley (1997) and Pawson (2008) for a fuller exposition of this argument, which
underlies the ‘realist’ approach to the synthesis of evaluation results.
21 The need to respond to the global financial crisis which broke in 2008 might also be mentioned here but, as we have stated, the economic impact was less than expected and the policy response relatively swift and decisive. (See Section 2.1 above and Sections 4.1 and 5.1.)
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 46
This has been a period of increasing scepticism over Budget Support from the side
of the Development Partners and increasing disappointment with the aspirations
of the Paris Declaration, with which Budget Support was associated. This has
reflected itself in pressures from headquarters to obtain quick results, to reduce
investment in institutional development and aid effectiveness work not
immediately related to quick results, and to be increasingly risk averse22.
It has been a period of increasing concentration of political power in Mozambique,
which has distanced government from citizens and reduced the need (and perhaps
the willingness) to listen to criticism and to alternative policy voices.
These factors are likely to have made more difficult the dialogue between
government and the Budget Support providers by reducing the commitment – on
both sides – to the joint process enshrined in the MoU for Budget Support.
Finally, the public policy agenda has become more technically and politically
demanding. Attention has had to focus not just on the attainment of economic
growth, the initiation of reforms, and the expansion of service coverage but also
on the stimulation of poverty reduction, the implementation and “roll-out” of
reforms, and the achievement of increased quality in service provision. At the
same time, the inherent constraints to effective public policy have been relatively
unchanged since the 2000-2005 period, notably the continuing high levels of social
deprivation, the inaccessibility of the poor and the low human resource and
capacity endowments within the public sector.
55. Taken together, these factors point to the need to moderate expectations regarding the
likely impact of Budget Support in the evaluation period. They constitute a context likely to
complicate the process of dialogue and constrain the ability to design and implement good
public policies and spending strategies. In short, they add up to a challenging context for
the successful implementation of Budget Support.
Prospects for the future
56. This evaluation seeks not only to assess the past but also to lay down recommendations
for the future. The assessment of future prospects will depend critically on political
developments during the 2014 election year but there are reasons to believe that the
future context for Budget Support should be better than it has been over 2005 to 2012,
although there remains at least one significant threat on the horizon:
Steady economic growth and fast increases in domestic revenue have created a degree
22 This was confirmed to us both in the focus group discussion with Heads of Cooperation, and by most
of the HoCs who were interviewed individually.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 47
of fiscal space, which Mozambique has never experienced before. While economic
growth has averaged 7.3 % per annum throughout the evaluation period, the revenue
to GDP ratio has increased from 15.3 % in 2006 to 23.3 % in 2012 (IMF, 2014). If
projections for continued economic and revenue growth are fulfilled, then domestic
revenue should exceed 25 % of GDP by 2018. This will permit Mozambique to
contemplate, from 2015 onwards, an ambitious programme of public spending
targeted on poverty reduction. In addition to expanded social spending and investment
in physical infrastructure as in the past, it will for the first time allow scope for social
transfers, targeted to directly address malnutrition, and other similar problems of
social deprivation. Continued or even expanded provision of Budget Support could
allow such a programme to be quickly scaled up, without serious risk to long-term
fiscal sustainability.
Due to the rise of the MDM as a viable third party, the 2014 elections promise to be
more tightly contested than in the past, potentially heralding a period of more
responsive government. Although Renamo refused to participate in the November
2013 municipal elections, MDM were able to compete with Frelimo in all 53
municipalities, winning the elections in three important cities – Beira, Quelimane and
Nampula, and securing a significant number of assembly seats in the key cities of
Maputo and Matola. With a further 9 months of preparation available, the October
2014 presidential and parliamentary elections promise to be more competitive than
the 2009 election. Frelimo as the incumbent government and the most established
party will clearly hold the upper hand but it will need to be more mindful of voters’
concerns than it has been in the past.
The increasing demand for evidence-based aid strategies, combined with the
availability of this evaluation creates a better framework for the design of effective
Budget Support to Mozambique. Led in large part by the efforts of research based
organisations like CGD and 3-i-e, DPs have been compelled to make increasing use of
evaluation studies, and in particular of impact evaluations. Different agencies have
responded to differing degrees to this trend but, in general, it has led to a higher level
of accountability and to a more careful use of evaluation evidence. Notwithstanding
the inevitable limitations of this evaluation, the mere existence of an up to date
evaluation should assist in the design of a more effective future Budget Support
package for Mozambique.
Mozambique’s forthcoming resource boom remains the biggest threat, opening
possibilities for unsustainable foreign borrowing, and creating a potential incentive for
less accountable and less transparent government. Resource extraction will not bring
significant domestic revenues to Mozambique before 2020 but it has already opened
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 48
up the opportunity for borrowing against these future earnings. The $850 million
recently borrowed by the Government for tuna fishing vessels for EMATUM and for
naval patrol vessels is an indication of the sort of commercial loans potentially
available, where a government guarantee is provided. However, at an interest rate of
8.5 %, this is relatively expensive financing and further borrowing at these levels could
generate an unsustainable debt burden, particularly if the resulting investments did not
themselves serve to generate income. At the same time, the investments associated
with the resource boom will generate significant opportunities for Mozambican
businesses in the construction industry and in the service activities, needed to support
these large-scale investments. There are powerful incentives for the political elite to
secure preferential access to these opportunities, and to design institutional
arrangements which avoid transparency and public accountability. To date, the work on
the design of the legal and fiscal infrastructure for the extraction industries has been
undertaken in close partnership with the IMF and has drawn on a wide range of
international advice. Efforts have also been made to reach EITI compliance. It is to be
hoped that this approach continues and that government leaders, officials, CSOs and
citizens can work together with their international partners to establish a mutually
beneficial set of arrangements for managing the resource boom.
3. Providing the means to implement policy: Budget Support inputs and direct outputs
57. This chapter addresses levels 1 and 2 of the evaluation framework, covering Budget
Support inputs and their direct outputs. Thus, it brings together the responses to EQs 1.1,
1.2, 2.1, 2.2 and 2.3,, which are also presented in tabular form with details of data sources
and their quality in Volume Two. Here, we synthesise these responses in order to address
the crosscutting question: “how successful has Budget Support been in providing the
Government with the necessary means to implement the national poverty reduction
strategy?” We consider four aspects:
What has been the scale of Budget Support, the relative predictability of its
disbursements, and its effects on the harmonisation and alignment of aid as a
whole?
What mechanisms for policy dialogue have been established through the Budget
Support framework at the aggregate and sector levels and how successful have
these been in promoting sustainable and effective processes of policy analysis and
development?
What technical assistance and capacity-building inputs have been provided?
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 49
How has the design of the Budget Support process been adapted to the changing
context, and with what effects?
58. The definition of budget support used by the evaluation follows the 2006 OECD-DAC
criteria which defines Budget Support as follows: “budget support is a method of financing
a partner country’s budget through a transfer of resources from an external financing
agency to the partner government’s national treasury. The funds thus transferred are
managed in accordance with the recipient’s budgetary procedures”.
59. Arrangements that do not rely entirely on government procedures for management of
funds are not considered budget support. During the evaluation period, Sector Budget
Support arrangements have been put in place by the EU to support the agriculture and
health sectors but these have not been managed by the government as Budget Support.
The SBS resources transferred to the treasury account have been assigned to the project
accounts, used for the management of the PROSAUDE and PROAGRI common basket funds.
Once in these project accounts, the management of these resources follows procedures,
which differ from national budgetary procedures in three important respects:
The detailed budget for the use of these resources – that is the breakdown of
spending by economic classifications and, where appropriate, programme
classifications – is not submitted to the National Assembly for approval but rather
to common basket fund committees, where external donors have a majority
representation.
The use of resources during budget execution does not follow the rules of internal
control laid down in the SISTAFE procedures: in particular, significant reallocations
of resources are made without approval of the Ministry of Finance or the National
Assembly, and substantial resources are used for the payment of salaries and
salary supplements, which would not be permissible under these rules.
The balances in these project accounts remain available for spending after the
close of the year, without their use having to be re-approved by the National
Assembly.
60. Given these significant differences, the evaluation team has classified these resources as
common basket funds (CBF) and not as sector budget support. However, the three major
CBF arrangements - PROAGRI, PROSAUDE and FASE - have provided substantial sector
funding, while contributing to capacity buidling and to policy dialogue at the sectoral level.
The relative influence of these arrangements in comparison with Budget Support has
therefore been a relevant question for the evaluation.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 50
3.1. Scale & Predictability of Budget Support, and its effects on aggregate Aid flows
Table 7: Budget Support Disbursements by Development Partner, 2004 - 2012
AGENCY (Millions of USD)
23
2004 2005 2006 2007 2008 2009 2010 2011 2012
GRANTS - - - - - - - - -
Germany 3.99 4.34 11.93 13.49 16.74 18.67 18.98 18.08 11.62
Austria - - - - 2.47 4.33 3.90 4.62 3.10
Belgium 3.67 2.04 3.54 4.08 4.67 - - - -
European Union 58.04 53.75 46.45 58.38 68.57 86.59 83.06 71.91 74.84
Canada - 2.03 1.90 4.26 7.57 9.89 14.21 16.03 14.84
Denmark - 9.54 2.49 19.20 10.55 8.51 7.34 12.96 10.69
Spain - 3.78 3.59 4.23 6.22 9.87 9.71 9.47 -
Finland 4.55 4.91 5.85 6.79 10.39 8.76 8.43 10.31 9.27
France 3.77 3.19 3.24 2.65 3.04 3.10 2.72 2.90 2.52
The Netherlands 17.16 14.01 21.24 24.16 26.23 25.15 21.68 25.83 11.24
Ireland 7.01 6.03 6.80 12.28 15.17 14.82 14.72 16.12 11.55
Italy 3.78 3.32 3.55 5.35 5.86 5.27 5.68 5.28 5.14
Norway 8.12 15.04 17.01 22.61 29.70 23.82 27.74 29.58 23.44
Portugal 1.48 1.53 1.41 1.49 1.51 1.60 1.93 - 3.90
United Kingdom 27.53 46.37 55.69 69.81 81.71 58.30 54.16 136.44 77.65
Sweden 12.53 17.95 23.92 44.45 56.50 40.12 44.42 51.35 43.40
Switzerland 7.20 7.88 6.33 6.95 7.08 6.64 6.91 4.18 7.30
SUB TOTAL 158.84 195.73 214.95 300.19 353.98 325.43 325.59 415.06 310.48
LOANS - - - - - - - - -
AfDB - - 58.08 30.29 32.31 30.28 30.05 28.20 30.32
World Bank - 65.91 56.48 69.37 72.09 78.72 88.77 92.90 108.07
SUB TOTAL - 65.91 114.55 99.66 104.40 109.00 118.82 121.10 138.39
TOTAL 158.84 261.64 329.50 399.84 458.38 434.44 444.40 536.16 448.86
Source: Direcção Nacional do Tesouro – Ministry of Finance.
61. In nominal US dollar terms, budget support has increased almost three-fold from 2004 to
2012: from just under US $ 160 million to just under US $ 450 million24. Budget Support
flows peaked in 2011 with total disbursements of US $ 536 million25 (Table 7).
62. In terms of fiscal significance, Budget Support has had considerable importance,
comprising some 18.8 % of public spending at its peak in 2006 (Table 8). Since then, it has
decreased steadily, reaching 9.4 % in 2012. However, within this period the growth of
domestic revenue was dramatic, from 11.22 % of GDP in 2004 to 23.3 % in 2012. With
GDP itself also expanding rapidly, the relative significance of Budget Support has been less
in recent years than in the period 2006 to 2010.
Table 8: Disbursements of Budget Support and their significance
2004 2005 2006 2007 2008 2009 2010 2011 2012
23 Mid-year exchange rate MZM-USD for every June 1: 23.12 (2004): 23.75 (2005): 26.78 (2006); 25.92
(2007); 24.10 (2008); 26.50 (2009); 33.75 (2010); 29.53 (2011); 27.93 (2012)
24 Figures presented here and in Table 7 have been calculated based on actual disbursements in MZM multiplied by its mid-year exchange rate (1st of June for each year).
25 This figure includes three payments by DFID which comprised part of a DFID BS operation for 2012, but whose transfers were made in 2011: 540,153,108.40 MZM on July 7, 2011; 507,750,428.91 MZM on September 21, 2011: and 501,217,718.40 on December 13, 2011. If these disbursements had been registered by the DNT within 2012, total disbursements for 2011 would have been US $ 483 million.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 51
Total Budget Support (Mill USD)
158.84 261.64 329.50 399.84 458.38 434.44 444.40 536.16 448.86
Budget Support as % GDP 2.9% 4.1% 4.9% 5.0% 4.6% 4.3% 4.8% 4.3% 3.1%
Budget Support as % Total Revenue
22.2% 30.5% 32.7% 31.4% 28.9% 24.3% 23.6% 19.6% 13.2%
BS as % Public Spending 11.2% 17.4% 18.8% 18.1% 16.6% 13.3% 15.0% 12.0% 9.4%
BS as % External resources (excluding
debt relief) 16.5% 30.5% 25.8% 34.2% 29.1% 21.4% 28.7% 24.4% 23.7%
Source: Ministry of Finance and BoM fiscal tables.
63. Over the 2005 – 2012 period, the proportion of Budget Support within total ODA has
represented on average just over 27 %, excluding debt relief, although this has declined in
2012 to 23.7 %. Thus, as a consequence of Budget Support, and the fact that most
common basket funds are on budget, a large proportion of total aid has been channeled
through the national budget. However, the proportion of aid on budget probably declined in
2012. Moreover, projects – virtually all of which are managed off-budget - have
represented a high proportion of ODA flows throughout the period (34% on average).
Figure 6: Importance in public spending of Budget Support and other aid modalities
Source: Ministry of Finance and BoM fiscal tables
The predictability of Budget Support flows
64. Annual predictability of budget support is significantly better compared with the two other
aid modalities (Basket funds and Projects). Actual disbursements of Budget Support have
generally been close to 100% of the planned disbursements, averaging 116% over the 8-
year period since 200426. This can be attributed largely to the success of the procedures
established in the MoU for estimating and reporting on the planned Budget Support
26 For Budget Support, data have been provided by the Direção Nacional do Tesouro - Contravalores of
MoF, who collect information from the CUT. Data regarding Basket funds and projects has been extracted from the Conta Geral do Estado. During the field mission, cross-checking of sources has demonstrated that these are more reliable sources of disbursement information than the ODAMOZ database.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 52
disbursements during the annual reviews and confirming projections prior to the start of
each fiscal year.
65. By contrast, the equivalent procedures for other modalities are less established and not so
uniformly respected by the various providers of projects and common basket funding.
Moreover, resource persons from both government and DPs agreed that the room for
applying disbursement conditionalities with projects and common basket funds is greater.
For these modalities, the variation in predictability has been more significant and
disbursements have usually been less than projected. This is particularly the case for
projects, which averaged disbursements of 56% of projections during the evaluation period
with a considerable variance around this mean. Basket fund disbursements were more
stable, but also averaged significantly below Budget Support (79% versus 116%), as a
percentage of projections.
66. The relative importance of Budget Support has decreased during the period in comparison
with the other two modalities. Thus, if we consider the overall predictability of the three
modalities together, it can be observed that although Budget support serves to make total
aid more predictable, its declining relative weight limits that influence (Figure 7).
Figure 7: Actual annual disbursements as a percentage of the planned disbursements: Overall predictability of aid (left); Individual detail for each modality (right)
Source: CGE (2005-2012) and REO (2012)
67. The lack of predictability of projects and common basket funds has seriously affected the
execution of the budget. The levels of execution of the external component of the
development budget (i.e. basket funds and projects) are dramatically lower than the
internal component and also much lower than average disbursement rates for Budget
Support. Running a projection scenario, in which external projects and basket funds are
executed at the same rate as internally funded development spending, the potential
0%10%20%30%40%50%60%70%80%90%
100%110%
2005 2006 2007 2008 2009 2010 2011 2012
ProjectsBasket FundsBudget SupportTotal Predictability of external sources (GBS+BF+Projects)
0%
20%
40%
60%
80%
100%
120%
140%
160%
180%
200%
2005 2006 2007 2008* 2009 2010 2011 2012
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 53
increment in public spending would have been equivalent to 5% of GDP in 2008 and 3% of
GDP in 2012 (sky blue area in the figure below).
Figure 8: The “lost” expenditure due to low execution rates of Common Basket Funds and externally financed Projects
Source: Fiscal tables from the Bank of Mozambique for years (2004-2011), fiscal table for 2012 from the IMF. Percentage execution of internal and external components from the CGE and REO for 2008 and 2012.
68. Budget Support disbursements are also projected by month. The timeliness of
disbursements within the year represents an important influence on treasury management
processes. For this reason, the MoU and many individual Budget Support agreements
emphasise that disbursements should be predictable, and ideally “front-loaded” within the
year. Over the evaluation period, an increasingly higher proportion of BS funds were
disbursed in the first two quarters. (Figure 9.)
Figure 9: Budget support disbursements “front-loaded” in the first two quarters
Source: CGE (2005-2011) and REO (2012)
69. From 2005 to 2009, the gap between quarterly projections and actual disbursements was
close to zero. Since 2010, differences have been higher, meaning that a greater proportion
of Budget Support disbursements are being delayed. Figure 10 shows the difference
between the projected Budget Support disbursements by quarter (as recorded by MoF -
DNT) and the actual disbursements in each quarter. When the difference is positive (as for
0%
5%
10%
15%
20%
25%
30%
35%
40%
2005 2006 2007 2008 2009 2010 2011 2012*
Composition of Central Government expenditure (% GDP)
'Public Service Delivery atfull potential (BasketFunds+ Projects)
External component ofdevelopment spending(actuals)
Internal component ofdevelopment spending(actuals)
Recurrent spending(actuals)
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
2004 2005 2006 2007 2008 2009 2010 2011 2012
1st Quarter 2nd Quarter Total 1st & 2nd Quater
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 54
the first quarter 2006), it means that the disbursements have been higher than
programmed. In general, we see that the differences are balanced out by the end of the
year, confirming the positive annual disbursement data reported above.
Figure 10: Differences between scheduled & actual GBS disbursements by Quarter (Millions of MZM)
Source: DNT – Ministry of Finance.
70. Field mission interviews suggest that the positive peak of 2006 is mainly explained by
shortcomings in the system of projection of disbursements, which were corrected in
subsequent years. Regarding the shortfalls in disbursements in the first two quarters of
2010, the first quarter of 2011 and the second quarter of 2012, documents and supporting
interviews confirm that these were the results of deliberate delays in disbursements by
some bi-lateral Budget Support providers, due to concerns over the ‘underlying principles’
for Budget Support having been potentially breached, through electoral irregularities in the
national elections of October 2009 (leading to the subsequent “crispação” with
Development Partners), health procurement irregularities in 2011 and education payroll
irregularities in 2012.
71. There is some controversy over whether these three suspensions were justified. The Budget
Support MoU of 200927 includes a commitment by DPs to “not interrupt in-year
disbursements unless there is a violation of the underlying principles”. In each of these
cases, there were grounds to suspect that a violation of the underlying principles might
occur but subsequent corrective actions by Government prevented a formal violation of
underlying principles from being declared. Government stakeholders maintain that the
“pre-emptive” suspension of Budget Support by certain partners was not needed, while
several DPs expressed a conviction that corrective actions would not have been taken or
would have been delayed in the absence of the decision to suspend disbursements. We
return to this point in Section 3.4
27 Point 25 of the Memorandum of Understanding (2009).
-3000
-2000
-1000
0
1000
2000
3000
4000
1Q
-05
2Q
-05
3Q
-05
4Q
-05
1Q
-06
2Q
-06
3Q
-06
4Q
-06
1Q
-07
2Q
-07
3Q
-07
4Q
-07
1Q
-08
2Q
-08
3Q
-08
4Q
-08
1Q
-09
2Q
-09
3Q
-09
4Q
-09
1Q
-10
2Q
-10
3Q
-10
4Q
-10
1Q
-11
2Q
-11
3Q
-11
4Q
-11
1Q
-12
2Q
-12
3Q
-12
4Q
-12
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 55
72. In practise, the financial costs of these temporary suspensions do not appear so great,
amounting to delays of 3-6 months in disbursements of amounts varying between US 50
and 100 million [i.e., approximately 3% (2010), 1.2% (2011) and 2.1% (2012) of public
spending]. The evaluation team were unable to discern any clear relationship between
Treasury Bill rates and delayed Budget Support disbursements in this period; nor were
there reports of damaging delays in authorisation of planned budget spending.
73. Thus, we conclude that the objectives of annual predictability and of “front-loading” of
disbursements were largely fulfilled during the evaluation period. Budget Support providers
were clearly conscious of the importance of these objectives and managed to maintain
good levels of predictability even through a period of difficult financial circumstances in
many donor countries. However, in-year predictability was compromised in 2010, 2011 and
2012 due to concerns over potential violations of underlying principles.
Evidence of positive effects of budget support on aid as a whole: harmonisation, alignment and use of country systems
74. How far has Budget Support had positive effects on the structure of aid as a whole? Two
direct effects envisaged from Budget Support are firstly an increase in the absolute and
relative level of aid managed through the national budget process (and more generally,
through country systems) and secondly an increase in the predictability of aid
disbursements. In addition, the Intervention Logic postulates that the harmonisation and
alignment of budget support inputs may contribute to the wider harmonisation and
alignment of external aid as a whole, with corresponding reductions in the relative
significance of transaction costs.
75. The importance of Budget Support within total ODA has decreased during the evaluation
period from over 30% to less than 25%. (See Table 8). In conjunction with the lower
predictability of projects and basket funds, this has meant that the predictability of overall
aid has declined over the evaluation period (See Figure 7, above.) Nevertheless, it is quite
conceivable that dialogue and management structures for Budget Support should exert a
positive influence on the management of aid, which is independent of the scale of Budget
Support.
76. The monitoring surveys of the Paris Declaration show a positive trend in the use of country
systems from 2005 to 2010. (Table 9.) The sources of data for these surveys may be
subject to bias, given that they are questionnaire-based, and there is also doubt over the
precise definitions of “use of country PFM systems” that have been applied28. However, the
28 The final evaluation of the Paris Declaration acknowledged these weaknesses in the data sources for
the 2008 and 2010 monitoring reports and drew its conclusions predominantly from a new set of Country and Donor reports, specially commissioned for the evaluation.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 56
PEFA assessments also report a small improvement over the evaluation period in scores
for total ODA managed through national systems, scoring “D” (2006), “D” (2008) and “C”
(2010). The data reported on use of country systems within the PAF indicators for the
Programme Aid Partners (PAPs), show some improvement from 2006 to 2009 but suggest
a decline thereafter, even though there are some issues of comparability across the years.
(See Table 9.)
Table 9: Trends reported in use of Country systems: Paris Declaration Monitoring Surveys and indicators in PAPs’ Performance Assessment Framework
PARIS DECLARATION MONITORING SURVEY 2005 2007 2010 Assessment
% of total ODA that use national PFM systems
36% 44% 47% Not achieved with progress
% of total ODA that use of national procurement systems
38% 54% 57% Not achieved with progress
Source: Paris Declaration Monitoring Survey.
ANNUAL REVIEW RESULTS (PAPs’ PAF) 2006 2009 2012 Trend
% use of national financial reporting systems
49% 68% 59% Not comparable series but always below the target
% use of national audit systems 37% 47% 44% Not comparable series but always below the target
% use of national procurement systems
52% 70% 57% Not comparable series but always below the target
Source: Budget Support Annual Reviews.
77. Notwithstanding the decline in its relative influence since 2009, we conclude that Budget
Support made a positive contribution to harmonization of aid. Several studies, notably the
Paris Declaration Monitoring Reports record that at the beginning of the period
Mozambique was frequently used as an example of donor harmonization. From the initial
conception of the Budget Support architecture, DPs understood that the GBS group should
lead harmonisation efforts across a much wider front than just Budget Support. Indeed,
Mozambique is unique among major Budget Support recipient countries in having
established explicit links between the management framework for Budget Support and the
wider goals of aid effectiveness.
Figure 11: Results of PAPs’ indicators within the PAF over 2005 – 2012
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 57
Source: Budget Support Annual Review reports
78. Performance against DPs’ aid effectiveness indicators was strong during the period 2007
to 2009, after a period of fine-tuning of indicators over 2005-2006. This reflected a
mutual effort to increase the quality of DP performance under the umbrella of the aid
effectiveness agenda. However, since 2009 a sharp drop in achieved indicators is evident,
culminating in 2012 when 22 indicators out of 23 were not achieved. (Figure 11.)
79. Aggregating the assessment scores for the PAPs’s indicators over the whole evaluation
period, 44% were assessed as not achieved, while 42 % were assessed as achieved or ‘not
achieved, with progress’. (Figure 12.) The indicators related to alignment and harmonisation
scored relatively better, with 48% assessed as achieved or ‘not achieved with progress’. As
noted above the decline in performance is particularly notable since 2010, although there
are some issues of comparability across the period due to changes in the indicators.
Figure 12: Aggregate assessment of PAPs’ performance indicators, 2005-212
Source: Own compilation based on PAF indicator scores from Annual Reviews.
80. Government and DP stakeholders agreed that the Budget Support dialogue around PAPs’
performance had fallen short of expectations since 2010. Among the reasons presented,
interviewed resource persons agreed on the following:
There were excessively ambitious targets established for PAPs’ indicators, which in
several cases were set higher than those in the Paris Declaration.
0
5
10
15
20
25
30
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2005 2006 2007 2008 2009 2010 2011 2012
Total number of indicators
Achieved
Not achieved but with progress
Not achieved
N/A
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 58
DPs had since 2010 shown less concern for aid effectiveness goals, which had
reduced the level of investment in aid effectiveness initiatives made at both
Headquarters and field office levels.
Government too needed to develop mechanisms to ensure a more systematic
follow-up on aid effectiveness issues, which would go beyond simple monitoring
and would also focus on how to facilitate use of country systems (e.g. through the
development of detailed guidelines) and increased alignment with government
policies and systems.
81. Nevertheless, the framework of mutual accountability incorporated within the PAF has
been highlighted as one of the most important gains in the relationship with Development
Partners. Government officials were unanimous in stressing the importance of this mutual
commitment to enhance ownership and to create a balanced relationship in the Budget
Support dialogue. Moreover, Mozambique has become an international reference point for
this mutual accountability framework.
82. There is evidence to suggest that the PAPs’ PAF is a useful mechanism to keep aid
effectiveness objectives on the agenda and to maintain some measure of mutual
accountability, even if the achievement of full horizontal accountability is unlikely, given
the inherent imbalances in the relationship. Stakeholders agreed that, if this mechanism is
to be retained, some revision of the design of the indicators would be necessary. This
should aim to reduce the overall number of indicators to a more manageable number, for
example no more than 12 indicators. Secondly, it should seek to set targets, which are
feasible to achieve, aiming for steady progress towards agreed goals, rather than
excessively ambitious short-term targets. In addition, a more structured follow-up by
Government and DPs would be necessary, perhaps focused around an agreed set of
procedural reforms and administrative steps – something akin to an “area for special
attention matrix” for aid effectiveness.
Influence of Budget Support on Transaction Costs
83. Evidence suggests a possible increase in overall transaction costs over the period of the
evaluation. This is partly the result of the decline in the importance relative to other
modalities of Budget Support – the modality reported to have the lowest transaction costs;
and partly a consequence of increased transaction costs within the Budget Support process.
The available evidence is indicative rather than conclusive, because it is based almost
exclusively on perceptions but all of these perceptions point in the same direction, with no
contradictory opinions having been identified. The evaluation team interviewed a range of
relevant government and DP stakeholders, including staff with a “central agency” (i.e MF/
MPD) perspective and staff with sector agency perspectives. They were asked to report
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 59
their perceptions on trends in transaction costs over time and by modality. Results were as
follows:
There was unanimous agreement that the transaction costs per unit of aid
remained lower for Budget Support than for other modalities.
Nevertheless, the transaction costs for Budget Support were perceived to have
increased up to 2010, and more or less remained at that peak since then,
prompting what was perceived as a certain level of disinvestment in the modality
from both Government and the DPs, with less staff time dedicated to dialogue and
to Budget Support-related analysis.
A reduction in the number of PAF indicators for Budget Support was noted.
Nevertheless, this was perceived to have been counterbalanced by increasingly
protracted discussions during the Annual Reviews, and increasing use of
alternative performance assessment criteria, such as “areas for special attention”.
Although there was a wider range of opinions on this issue, the majority view was
that the reduced PAF had not yet succeeded in reducing transaction costs.
The balance of opinion was that transaction costs for Basket Funds had remained
steady over time, as management and coordination procedures had already been
streamlined before the evaluation period.
No discernible change was reported in the transaction costs of projects, but
government resource persons, at central and sectoral levels, emphasized their
dissatisfaction with the excessively high transaction costs of the project modality,
as well as their concerns over the predictability of disbursements. (See Figure 7
above.)
Regarding the large vertical projects in the Health sector, Government found the
transaction costs of working with the Global Fund very high, and it is widely
accepted that the transaction costs of the other vertical funds (PEPFAR, etc) are
relatively high since they include many parallel structures.
Table 10: Trends in PAPs performance indicators related to aid transaction costs
2006 2009 2012 Trend
% Programmatic Aid 70% 67% 63% Negative, and always below the targets
Total number of missions 203 167 158 Positive but not achieving the target
% of joint missions 10% 25% 26% Positive but not achieving the target
% of ODA using national financial reporting systems
49% 68% 59% Not comparable data set, but always below the target
% of ODA using national audit systems
37% 47% 44% Not comparable data set, but always below the target
% of ODA using national procurement systems
52% 70% 57% Not comparable but always below the target
Total number of Project Implementation Units (PIU)
21 20 26 Negative, and always below the targets
Source: Budget Support Annual Reviews
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 60
84. These perceptions are supported both by the Paris Declaration monitoring reports29 and by
the trends in PAPs’ performance indicators related to aid transaction costs (Table 10).
Although modest progress is reported in some areas, such as numbers of missions,
numbers of joint missions and use of country systems, none of the targets in these areas
have been accomplished as expected. Moreover, performance has deteriorated on the use
of programmatic aid and on the use of Project Implementation Units (PIUs).
3.2. Influence of Budget Support on policy dialogue processes
85. This section examines the extent to which budget support has contributed to sustainable
improvements in the processes and methods of policy development and monitoring - in
terms of institutionalising efficient frameworks, which involve relevant stakeholders, focus
on strategy and results and help to resolve policy dilemmas. We first examine the nature
and quality of the policy dialogue frameworks at national and sectoral levels and their
evolution over the evaluation period. We then consider the evidence that GoM structures
for policy dialogue have been influenced in a positive and sustainable way by budget
support processes, considering also the potential counterfactual that the positive
developments have been driven by SWAP frameworks or other non-Budget Support factors.
This latter analysis is focused on the agriculture, health and education sectors, where the
PROAGRI, PROSAUDE and FASE common funds are prominent.
Framework for policy dialogue at national level
86. The G19 is widely acknowledged as the most influential group in Mozambique due to the
coordination structure related to it and the access to high-level dialogue with the GoM,
which it provides. Although the DPG consists of the heads of mission of virtually all
multilateral agencies and bilateral donors, including the G19 and non-GBS donors, the
group is primarily for information sharing within the development community and does not
interact as a group with the GoM.
29 See for example the Paris Declaration Monitoring Survey (2011): “There seems to be a certain
weariness with the lengthy and cumbersome process of coordination, and the associated transaction costs.”
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 61
Figure 13: Budget Support Coordination and Dialogue Mechanisms
Source: MPD 2013
87. The structure for policy dialogue and assessment of performance in GBS was formalised in
the 2004 Memorandum of Understanding and retained within the 2009 MoU. The principal
mechanisms for GBS coordination and dialogue are illustrated in Figure 13 and are as
follows:
Political dialogue: takes place at the Head of Mission/ Ministerial level.
The Joint Steering Committee/ Management Group is comprised of the national
directors of the MPD, MoF, MINEC and the Central Bank with the MPD acting as
chair. The PAPs are represented through the “Troika-Plus” which comprises three
members from the PAPs, who rotate every three years with one new member
joining each year. There are in addition two permanent members (the World Bank
and the EU).
Sector and thematic groups report to the management group with the GoM
represented through sector ministry representatives and DPs through their
technical specialists. CSO representatives also sit on a number of sector and
thematic groups.
88. For the sector and thematic dialogue, a five pillar framework, matching the structure of the
PARPA I, was originally established, with pillars covering i) poverty and macroeconomic
management; ii) governance; iii) economic development and; iv) human capital, in addition
to a pillar on cross-cutting issues. There were also a series of working groups beneath each
pillar, with donor focal points and GoM group leaders assigned to each working group. (See
Table 11.)
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 62
Table 11: Pillars & Working Groups for PARPA/ Budget Support monitoring, 2004-2009
Thematic Pillars Working Groups
1) Poverty and
Macroeconomic Management
Growth and macroeconomic stability
Poverty analysis and monitoring systems
Public financial management consisting of 5 subgroups:
budget analysis group, tax reform, procurement reform,
SISTAFE and Audit
2) Governance Public sector reform, Decentralisation, Justice sector
reform
3) Economic Development Financial sector
Private sector
Agriculture and rural development
Road sector (including telecoms and railways)
Energy sector
4) Human Capital Health
HIV/AIDs
Education
Water and sanitation
Cross-cutting Gender, Environment, HIV/AIDs
MoU Implementation PAPs assessment
Source: 2004 MoU for Budget Support
89. The working group structure changed in 2011 with the development of the PARP. This
established a new structure that was more cross-cutting and thematic. It increased the
number of pillars to six, including one on cross-cutting issues, and replaced the working
groups with a series of thematic groups and sector groups. This new structure is shown in
Table 12.
90. Stakeholders consider that, in the earlier years of budget support, the structures for policy
dialogue between the GoM and the G19 worked well, with dialogue reported by both the
GoM and donors as being reasonably effective. More recently there is a general perception
that there has been a deterioration in the quality of the dialogue. In part, this deterioration
is attributed to the changes in the structure of the pillars and working groups introduced in
2011 - a structure which is not yet fully functional and which is perceived, particularly by
the DPs, as excessively cumbersome. In part, it is attributed to a decrease in the
commitment of both the Government and the DPs to resourcing the G19 agenda.
Table 12: PARP/ Budget Support Working group structure 2011-2014
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 63
Source: MPD 2013
91. DP stakeholders report that the post-2011 framework for policy dialogue structure is not
yet fully operational and that the sequencing and links between the thematic groups and
working groups have in practice been weak. There has also been a lack of clarity over who
should attend these groups, and over which groups should be regular policy dialogue
groups and which need only convene for the Annual Reviews. In short, more operational
guidance from Government is seen to be needed in order to clarify the functioning of the
proposed structure.
92. It is also reported that there is significant variability in the quality of the working groups.
This is particularly because there has been difficulty experienced by the DPs in staffing
some of the most important groups. There have been long gaps in the nomination of DP
leads in key G19 working groups such as the Governance Pillar, the Poverty Analysis and
Monitoring systems group, the Macroeconomic Pillar and the Budget Analysis Group.
The Performance Assessment Framework (PAF)
93. The Government’s Performance Assessment Framework (PAF) is a central element of the
dialogue structure. It is a multi-annual matrix of policy targets, derived from the
government’s poverty reduction strategies - PARPA I, PARPA II and PARP. It is the focus of
discussions at the Annual and Half-yearly Reviews and together with the assessment of
underlying principles is the principal basis for DPs’ decisions on disbursement of Budget
Support. The preparation of an effective PAF is a challenging task, given that it strives to
capture within a manageable and measureable set of annual indicators, a strategic and
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 64
relevant way of measuring progress towards the achievement of the poverty reduction
strategy, whilst also achieving a consensus between government and 19 DPs on the chosen
set of indicators. Many countries have struggled to agree on a PAF that could fulfil all of
these objectives30.
Table 13: Total numbers of PAF indicators per year by Sector/ Thematic Area
Row Labels 2005 2006 2007 2008 2009 2010 2011 2012 Total %
Public Financial Management 7 9 8 8 8 2 8 7 57 17%
Justice sector reform 11 11 5 5 5 5 3 3 48 14%
Decentralization 3 2 2 2 2 1 2 2 16 5%
Public sector reform 1 3 3 2 2 2 2 1 16 5%
Agriculture and Rural Dev. 6 4 5 5 5 6 7 8 46 14%
Health 6 6 6 6 6 6 3 3 42 12%
Education 4 4 3 3 3 3 4 3 27 8%
Macroeconomic & financial Policies 9 8 1 1 1 1 2 23 7%
Private sector 2 2 2 2 2 2 2 3 17 5%
Water and Sanitation 2 3 1 1 1 2 2 2 14 4%
Financial Sector 3 2 2 2 1 10 3%
Roads 3 1 1 1 1 1 1 9 3%
Social Action 1 1 1 1 1 1 6 2%
Gender 1 1 1 1 1 5 1%
Energy 1 1 1 1 4 1%
Total Government 54 54 42 41 41 35 38 35 340
Alignment and Harmonization 12 8 7 7 7 9 10 7 67 39%
Predictability 5 4 5 5 4 7 8 10 48 28%
Capacity Strengthening 2 4 6 5 5 5 6 3 36 21%
Portfolio Composition 2 2 2 3 3 3 3 3 21 12%
Total DPs 21 18 20 20 19 24 27 23 172
Total (Gov. + DPs) 75 72 62 61 60 59 65 58 512 (1) Justice, Decentralization and Public Sector Reform are considered as Good Governance. (2) Figures show percentage by category for the Government PAF and the PAPs’ PAF taken separately. Source: Own compilation based on assessments in Annual Review reports
94. How do we rank the quality of the Mozambican PAF? Evidence points to the following
strengths:
The PAF is closely aligned to the objectives and targets of the poverty reduction
strategy. Taken together, Public Financial Management, Good Governance,
Agriculture, Health and Education represent 74% of the total indicators scored.
(Table 13). Although the number of PAF indicators has fallen during the period, the
weight of these five areas has remained reasonably steady. Among them, Public
Financial Management issues have been the most targeted area (17%).
The choice of indicators included in the PAF is strategic, in the sense of including
issues of importance, which cannot be readily monitored and discussed in other
fora. For example, the limited number of sectoral indicators in the PAF reflects the
fact that there are sectoral working groups where it is more appropriate to give
30 The case of Tanzania is an interesting case in point. See Lawson et al, 2013.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 65
detailed attention to sectoral objectives and targets. Similarly, the reduced
attention to macroeconomic issues reflects the fact that they are being addressed
through the monitoring of the IMF Policy Support Instrument (PSI).
The number of indicators in the PAF has been reduced from 54 to 35. As we have
noted, it is not generally considered that this has reduced transaction costs to date
because the number of issues addressed in other ways (as “areas for special
attention”) has increased. Nevertheless, the move to a more streamlined PAF
should in the medium term restrain transaction costs..
Indicators are for the most part measurable, as may be seen from the relatively
small number (3-8 %), which are annually classified as “Not Assessed” (N/A).
(Figure 14)
The level of ambition of the targets for the PAF indicators looks broadly right. On
average, slightly over 50 % of targets are assessed as achieved, with a further 30
% assessed as “not achieved but with progress”. (Figures 14 & 15.)
Figure 14: Performance of Government PAF indicators, 2005 - 2012
Note: N/A = Not Assessed.
Figure 15: Performance of Government PAF indicators by sector/ theme
Source: Own compilation from assessments in Annual Review reports.
95. Based on these five criteria, our assessment of the Government PAF is that it fulfils its
intended purpose reasonably effectively and compares well with the PAFs used in other
Budget Support recipient countries. In stating this, we acknowledge that all indicator-based
0
10
20
30
40
50
60
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2005 2006 2007 2008 2009 2010 2011 2012
Total number of indicators
Achieved
Not achieved but with progress
Not achieved
N/A
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 66
assessments of public sector performance have at least two inherent limitations. Firstly,
they must focus on what is measureable, which is not always the same as what is
important, and, secondly, the adoption of indicators as policy targets almost inevitably
creates perverse incentives in the collection and reporting of data and/or in the process of
policy implementation31. Indicator-based assessments – such as PAFs - should therefore be
used as an overall guide to progress and should not be the sole method of assessment.
Complementary frameworks for Dialogue & Performance Assessment
96. With the PAF essentially established as a mechanism to monitor progress towards results
indicators linked to the PARPA/ PARP, its effectiveness as the core of an overall system of
dialogue must depend significantly on the quality of the complementary frameworks, which
feed into this central structure. At present, there are three policy dialogue frameworks
additional to the PAF: 1) the “areas for special attention” which define policy actions to be
taken in high priority cross-cutting reform areas, such as payroll reform; 2) the IMF’s Policy
Support Instrument, addressing macroeconomic and structural issues; and 3) sectoral
performance matrices for the main sectors. We briefly assess each of these in turn.
97. The “areas for special attention” essentially comprise matrices of policy actions and
administrative reforms agreed between Government and the G-19, in order to address
specifically identified policy problems, which are not adequately addressed through existing
sectoral strategies and reform programmes. They have been used in an “ad hoc” manner to
address gaps, which have been identified in the normal policy dialogue process. The
Governance Action Plan (GAP), developed in 2010 by the Government in response to the
concerns expressed by the G-19 over electoral fraud and over the weakness of anti-
corruption measures was apparently the first such “area for special attention”. More
recently, payroll reform has also been adopted as such, following revelations in late 2011
of payroll fraud in the education sector. In these two cases, Government and DP
stakeholders recognised the usefulness of these matrices of policy actions and their value
as a complement to the broader outcome targets captured in the PAF. However,
stakeholders did express concerns about the potential proliferation of such matrices and
the related transaction costs. There is also a doubt over how well these areas for special
attention are coordinated and integrated with the monitoring of the PAF and with the wider
31 To be specific, the use of an indicator as a target may generate incentives to suppress the
dissemination of alternative indicators, which might cast doubt on the degree of progress being achieved or even to “doctor” the data related to the indicator to be reported. It may also encourage attention to those aspects of reform implementation most closely linked to the indicator itself, at the expense of complementary actions, also required for progress to be sustainable. “Goodhart’s Law” essentially states that these perverse incentives are inevitable whenever a performance indicator is adopted as a policy target.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 67
structures of Budget Support dialogue.32
98. The IMF’s Policy Support Instrument (PSI) is the chief mechanism utilised by the
Government for defining targets for macroeconomic policy and for related structural
reforms. It is managed directly by the IMF in conjunction with the Mozambican authorities
but there are structured points of interaction between the IMF and the G-19 at the time of
PSI missions. As a mechanism for setting macroeconomic policy targets, defining actions to
achieve those targets and measuring progress, the evidence suggests that the PSI
mechanism is highly effective.
99. A notable point of comparison with the PAF is that, although the PSI is focused on a small
set of quantitative and structural benchmarks, which are formally monitored, there is an
in-depth analysis of the macroeconomic and other variables related to these benchmarks
and a policy discussion, which is much more wide-ranging than these benchmarks. By
contrast, in several areas of the PAF, evidence suggest that the discussion at the time of
the Annual Review does not go much beyond the indicators themselves: there is neither
sufficient analysis to assess why or why not indicators have been met, nor sufficient
breadth of discussion to properly assess the policy implications33.
100. In principle, this more in-depth analysis and more wide-ranging discussion should be
provided through the sectoral and thematic working groups but the evidence suggests that
the quality of these groups is highly variable, which impacts directly on the quality of their
inputs into the PAF Annual Review. We examine below the functioning of dialogue in three
sectoral structures – for health, education and agriculture.
Sectoral and Budget Support dialogues in Health
101. A Code of Conduct signed in 2000 (revised in 2003) sets out the basic rules of
engagement between the Ministry of Health (MoH) and its partners. There is also a
Memorandum of Understanding (MOU) which sets out working arrangements that enable
structured dialogue and consensus building between MoH and DPs.
102. The SWAp is long established – preceding GBS at the national level, and this structure
has continued throughout the period under review. From 2007 the health sector has
evaluated its performance based upon a sector PAF. In 2012 the PAF comprised 35
32 For example, for the purpose of this evaluation, documentation on these areas for special attention
was not available through the secretariat of the G-19 and had to be obtained from other sources. Thus, there may be other areas for special attention, of which the evaluation team were not made aware.
33 Clearly, the quality of the dialogue in the PSI process is enhanced by both the volume and the technical quality of the inputs provided by the IMF and by the Mozambican Authorities. The capability to provide equivalent inputs across the range of sectoral and thematic areas captured within the PAF does not exist and could not easily be created but there is certainly a need to strengthen the quality of research and analysis. We return to this point below.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 68
indicators. Disbursements are based on overall “satisfactory performance” of the sector
against the Health PAF, including performance in financial management. Assessment takes
place through the annual joint reviews (ACA). The 2010 ACA was the first to be based on
the provisions of PROSAUDE II.
103. Although the health sector dialogue process has its own history, energy and
momentum, the GBS dialogue process and the health sector dialogue processes are
mutually supportive. The main formal links are two. First, the GBS dialogue is based upon
the strategies of the PARP (and PARPA II prior to 2011). Within this, health has been
designated as a priority sector and within the national dialogue process, the percentage of
funds allocated to priority sectors is tracked, the level of budget execution in health is
monitored, and notable developments within the sector are documented and discussed. The
profile of the sector is therefore raised by the national level dialogue, created by the GBS
framework.
104. Second, a number of indicators from the health PAF are elevated to the national level
PAF, and these are discussed within the GBS dialogue. At present there are three such
indicators, which are: (i) rate of coverage for institutional deliveries; (ii) number and
percentage of HIV+ pregnant women and newborns receiving PMTCT prophylaxis; and (iii)
the ratio of health workers in the areas of medicine, nursing and maternal/child health per
100,000 inhabitants. These indicators were acknowledged in focus groups to be important
and highly relevant. Moreover, the team were informed by some health partners that the
selection of an indicator for the GBS PAF was useful in advocacy: in bringing central level
attention to bear upon issues of sectoral concern.
105. The indicators used in the health PAF over the period under review have been relatively
consistent, but this is not true for the health indicators selected for the GBS PAF. They have
varied in number. There were 6 health indicators in the GBS PAF from 2005 to 2010 and in
this period 14 different indicators were used, many for only one year before being dropped.
In 2011 and 2012 there are only 3 health indicators in the GBS PAF. The steadiest
indicators are rate of coverage for institutional deliveries (seven years out of eight) and
the number and percentage of HIV+ pregnant women and newborns receiving PMTCT
prophylaxis (all years). During the period these indicators have both improved substantially.
The institutional births ratio rose from 49% to 64% over the period, while the PMTCT
prophylaxis covered 84% of births to HIV+ women in 2012 as against only 6% in 2005.
However, many factors are involved in the movement of these indicators, and it is not
possible to say to what extent these areas benefited from their continuous presence in the
GBS PAF.
106. In addition to these formal links between GBS and the health sector dialogue there are
other links less easy to define. In particular health advisors raise issues with their HOMS,
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 69
and health issues are also raised with HOCS who may take them to the G-19. They may use
other fora strategically to push key health issues. In addition, it is likely that health has
benefited from the cross-cutting issues debated and promoted at the G19, such as support
for decentralization, the strengthening of PFM systems and the extension of the e-SISTAFE
system into districts.
107. Another factor is that the health sector dialogue extends beyond the formal dialogue
structures defined in the MoU. Major vertical projects such as PEPFAR, Global Fund,
FORSASS and others all have their own dialogue arrangements and regular meetings with
government. PEPFAR is present in all provinces, and has the opportunity for continuous
dialogue at provincial level, which may over time influence policy. It is significant that the
total amount of budgeted funding for vertical funds from the US Government alone
exceeds $300 million in 2013, which compares with approximately $450 million per annum
of GBS. The relationship between influence and funding levels is not quantifiable, but
typically major funders expect to engage fully and robustly with government agendas,
especially in their areas of support.
108. In conclusion, in the health sector there is no doubt that GBS dialogue and sector
dialogue are mutually supportive and informants suggest that national level dialogue
associated with budget support has provided additional weight in key policy areas.
However, the evolution of health policy dialogue has been driven primarily by SWAP
frameworks and reflects the ebbs and flows in the quality of dialogue between the
Ministry of Heath and its PROSAUDE partners. The influence of GBS structures is certainly
modest and, moreover, works alongside other competing sources of influence such as the
vertical programmes.
Sectoral and Budget Support dialogues in Education
109. In the education sector, clear structures exist for policy dialogue consisting of a series
of consultative and decision making structures. Over time the structures have matured to
better fit with national dialogue processes (including GBS structures). These structures are
well aligned with the policy priorities in the Education Plan. Government informants
unanimously underscore the importance of the policy dialogue and its improved quality.
Donor respondents acknowledge the added value of the structures, and the improvements
over time, but consider the sector structures burdensome. This is in part because of a
reducing staff contingent on the donor side.
110. The basic structure for education sector dialogue was established in 2002 at the time
that FASE (Fundo de Apoio ao Sector de Educação – Education Sector Support Fund) was
put in place. It continues to exist today, with minor modifications, mostly related to
frequency and duration of meetings. Memoranda of Understanding (MoU) have governed
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 70
the structures and processes. There have been three MoUs (2002 – signed by 5 partners;
2006 – signed by 10 partners; and 2012, signed by 9 partners). In 2008 an addendum to
the second MoU was made to accommodate the procurement requirements of the World
Bank, so as to channel FTI (Fast Track Initiative) funding to the sector through FASE. The
policy dialogue and monitoring takes place through a structured set of fora, which meet at
fixed times in the year.
111. Documentation and interviews with key informants highlight that over time the
dialogue process has matured and become more formalized. Measures taken have sought
to streamline both processes and results and have included changes in the timing of the
Annual Review to better align with the planning and budgeting cycle of the GoM. As the
systems for disbursement and reporting have become stronger, there has been a shift from
a concentration on process issues related mainly to financial management and reporting,
to discussion of policy implementation.
112. The policy discussion process has been influenced by global commitments and
initiatives. The FTI has been particularly important: “FTI supported existing government
education policies and had a catalytic effect on policy discussions, drawing attention to
areas that were not receiving sufficient attention” (Bartholomew et al., 2009). This included
the need for new, more sustainable policies in teacher education and school construction
but also attention to gender equity and to teacher salaries (Takala, 2008).
113. A recognition of the importance of linking the sector processes to GBS processes
resulted in a fine-tuning of the structures and timing of major policy dialogue events. In
addition, the sector has chosen to use in a strategic way the three education sector
indicators included in the GBS PAF, so as to bring national level attention to the issues of
importance to the education sector:
Increasing the number of teachers has been the most prominent issue over the
period and a policy priority in the two most recent education sector plans. The MoE
has effectively used the PAF indicator on the pupil-teacher ratio, to argue with the
MoF for increased funding for teacher’s salaries.
The shift from monitoring Gross and Net Enrolment Rates to explicitly monitoring
right age enrolment in year 1 has helped to bring attention to the strong link
between improved retention rates/ lower drop-out rates and right age enrolment in
Year 1.
The focus on decentralization has brought with it a series of decisions around the
need for better monitoring of provinces and districts and improved support for
capacity development, especially at the district level.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 71
A more recent policy discussion has been around learning achievement – i.e. the
quality dimension of education. On-going work at the time of this evaluation
related to developing a learning achievement baseline in primary education which
will be one of the three indicators to be monitored at the level of the G19 GBS
dialogue.
114. Thus, in the education sector too a robust framework of policy dialogue has been built
up drawing on the SWAP structures, which in fact preceded GBS at the national level.
Arrangements for sector policy dialogue have matured and improved over the years,
becoming more strategic, more policy focused and less process focused. This was
essentially an autonomous process, to which the GBS process contributed little if at all.
115. In more recent years, sector stakeholders have learned – just as in health - to use the
national level dialogue associated with budget support to lend additional weight in key
policy areas, notably the need to reduce pupil-teacher ratios through teacher expansion, to
introduce a learning achievement baseline so as to monitor the quality of education more
effectively, to manage the process of budget decentralisation more effectively, and to give
increasing attention not just to Gross Enrolment Rates but more specifically to right age
enrolment in year 1.
Sectoral and Budget Support dialogues in Agriculture
116. The mechanisms for policy dialogue between GoM and PROAGRI donors have remained
more or less the same during the evaluation period. Since 1999 two annual review
meetings between the GoM and donors have taken place. The first - the Agriculture Joint
Review Meeting - occurs in the second quarter of each year and is focused on discussion of
the Annual PROAGRI Report. At the same time a draft annual plan for the following year is
presented and donors announce their planned commitments for the coming year in the
Harmonisation Report. The second meeting - the Agriculture Mid-year Review Meeting,
attended by donors, MINAG, MPD and MoF, is held in the third quarter of each year and
donors and government then confirm their respective contributions to PROAGRI, based on
the annual funding needs. The PROAGRI Working Group, comprising representatives from
MINAG and donors, meets at least once a month and is the main fora for dialogue on
operational matters.
117. Higher level policy dialogue takes place in the Fórum de Concertação which meets
every quarter and is the principal forum for dialogue on policy and strategic direction for
PROAGRI II. The Minister, Vice-Minister or Permanent Secretary of MINAG chairs the forum
and it comprises senior staff members of the MPD and MoF, donor representatives and
senior staff. However, more recently policy has become more fragmented, as the
government has developed policies outside the policy framework supported by PROAGRI
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 72
and has allocated significant resources to these processes, notably to the development of
the Programa Nacional de Investimento do Sector Agrário (PNISA – the National Agriculture
Sector Investment Plan), linked to the Comprehensive Africa Agriculture Development
Programme (CAADP), initiated under the auspices of NEPAD.
118. The conditionality framework for PROAGRI I based disbursements on planning and
financial management procedures, such as quarterly disbursement forecasts, a successful
annual review and audit. This led to PROAGRI I being criticized for being too concerned with
processes rather than results. The 2007 MoU therefore introduced a performance
assessment framework, with a matrix of agriculture sector performance indicators.
119. At this stage, the GBS PAF included five indicators for the agriculture sector, focusing
on public extension, irrigation and land registration. However, the 5 agriculture sector
indicators included in the GBS PAF were considered insufficient to assess performance of
MINAG and PROAGRI and a PROAGRI Performance Assessment Matrix, with twenty three
indicators was developed and annual targets established. These indicators and targets are
reviewed annually by GoM and donors in the Agriculture Mid-year Review, which is part of
the broader GBS review.
120. As Cabral (2009) notes, the process of developing the PROAGRI PAF was not ideal,
which impacted on the quality of the matrix developed. The process was essentially driven
by donors on an ad hoc basis, through individual consultation with different parts of
MINAG. This has led to a disconnect between MINAG annual activity plans and the
indicators and targets in the matrix and as a result, the PROAGRI PAF is not taken seriously
by MINAG. The weaknesses of M&E systems are also still a major concern, while there has
been little involvement of non-state actors in policy dialogue and monitoring of PROAGRI
performance.
121. The link between the PROAGRI performance matrix and the agriculture indicators
included in the GBS PAF is less clear for agriculture than it is for health and education. The
number of indicators has increased gradually over time, reaching 8 in 2012 (See Figures
13, 14 & 15, above), although the focus on irrigation, land registration and public extension
has largely remained. In general, the assessment of these indicators has been reasonably
positive: over 2005 to 2012, 52% have been assessed as ‘achieved’, 24 % ‘not achieved
but with progress’ and 22 % ‘not achieved’. However, there is no clear sense that indicators
are being deliberately selected in order to increase the administrative attention given to
their implementation, and as we note in our analysis of agricultural policy and spending in
chapter 4, the indicators bear little relation to the policy priorities emerging from research
and evaluation. In particular, none of the 20 indicators used over 2005-2012 related to the
degree of access or use of fertilisers, improved seeds or pesticides, despite the fact that
agricultural survey data point clearly to input use as the principal determinant of
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 73
agricultural productivity, especially for smallholders. (See Chapters 4.4 & 6.)
122. To summarise, agriculture sector dialogue is based upon a set of well established
structures, which were established initially through PROAGRI and later linked to the GBS
PAF process. However, the functioning of these processes reveals a number of weaknesses,
which the link to the GBS process does not appear to have mitigated in any significant way,
notably:
The fragmentation of the national structures for agriculture sector policy making;
The weaknesses of monitoring and evaluation systems;
The apparent disconnect between the policy targets in the PROAGRI assessment
matrix and the activity plans and budgets of the Ministry of Agriculture; and
The apparent disconnect between the agriculture sector policy targets in the GBS
PAF and the policy priorities emerging from agricultural research and evaluation.
Influence of Budget Support on Policy Dialogue Processes: Conclusions
123. Prior to the evaluation period, Mozambique had already established the core essentials
of a structured, harmonised assessment and disbursement process for Budget Support,
linked directly to the achievement of the targets in the national poverty reduction strategy.
This system has been preserved and refined. An integrated annual review calendar
remains in place, clearly linked to a network of sectoral and thematic policy processes. The
proceedings of the annual review process are open to a wide range of stakeholders, their
results are published and are generally the subject of press releases.
124. In 2011, a number of changes to the framework for policy dialogue were introduced to
fit with the new PARP structure, which adopted a more thematic structure, designed to
encourage more inter-ministerial cooperation around a more cross-cutting, thematic
dialogue. Whilst the rationale for the new structure is clear, it seems that, in the short
term, its introduction has met with some “teething problems” and generated additional
transaction costs. Some simplification of the structure may perhaps be needed, in addition
to further guidance on how it is expected to work.
125. The dialogue framework was centred around two Performance Assessment
Frameworks (PAFs) one for Government and one for the Programme Aid Partners (PAPs).
There is evidence to suggest that the PAPs’ PAF is a useful mechanism to keep aid
effectiveness objectives on the agenda and to maintain some measure of mutual
accountability. However, stakeholders agreed that, if this mechanism is to be retained,
some revision of the design of the indicators will be necessary. This should aim to reduce
the overall number of indicators to a more manageable number, for example no more than
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 74
12 indicators. Secondly, it should seek to set targets, which are feasible to achieve, aiming
for steady progress towards agreed goals, rather than excessively ambitious short-term
targets. In addition, it should be linked to a more formalised framework for follow-up,
which could potentially be structured as a joint GoM-PAPs matrix of administrative,
procedural and policy actions, akin to an ‘area for special attention’ for aid efffectiveness.
126. Our assessment of the Government PAF is that it fulfils its intended purpose
reasonably effectively and compares well with the PAFs used in other Budget Support
recipient countries. Notably, it provides a strategic set of measureable targets, which have
worked reasonably well as a focus for dialogue. However, like all indicator-based
assessments of public sector performance the PAF has inherent limitations, which suggests
that it should be used as an overall guide to progress and should be neither the sole
method of assessment nor the sole locus of dialogue.
127. In this respect, the decision by Government and its Partners to assess the progress of
certain governance and public sector reforms through agreed matrices of policy actions,
described as ‘areas for special attention’ seems sensible, so long as this does not lead to a
proliferation of ‘ad hoc’ assessment frameworks. It would therefore be advisable to
formalise this approach so as to prevent such proliferation and so as to ensure that these
frameworks are recognised as legitimate, and are managed in a transparent and objective
manner, clearly integrated with the PAF and the wider GBS dialogue process.
128. In stating that the PAF has functioned in a reasonably effective manner, we also
acknowledge that there are important policy objectives, related notably to agricultural
development and to poverty reduction where the Budget Support dialogue, focused on the
PAF, has failed to generate results. Some might argue that a more “strategic” selection of
indicators within the PAF would have generated better results in these important areas.
Yet, the PAF must necessarily reflect the nature of a given set of government policies:
setting targets for economic or social objectives, for which there is no clearly defined
policy would simply result in low scoring annual assessments, while potentially failing to
give credit and attention to other areas of policy achieving greater success. In this sense,
we would assert that the PAF can only build upon the existing body of policies, helping to
give focus to those policies, to establish sensible implementation targets and to provide a
public forum for their assessment. If there are economic or social objectives for which well
defined policies are missing, the PAF is unlikely to be the right instrument to generate an
appropriate policy analysis and development process.
129. In our opinion, it is the policy processes at sector and thematic levels, which should
generate the analysis, the testing of ideas and the debate from which appropriate policy
solutions should emerge. National policy structures, such as the PARPA/ PARP should
provide direction in relation to overall priorities but it is at the sectoral and thematic level
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 75
that the detailed analysis of problems and constraints should take place, leading to the
development and consideration of different policy options, from which an implementation
strategy can emerge. The PAF can then consolidate policy targets and provide a multi-
stakeholder process for assessing progress and adjusting implementation plans as
appropriate.
130. Our assessment of the policy dialogue process in Mozambique suggests that the main
structural weakness lies precisely within the policy development process, which should be
situated at the sector and thematic level. There are several problems of strategic
importance to the PARP agenda – such as low productivity in agriculture, high levels of
malnutrition, a poor enabling environment for doing business - where effective policy
responses have simply not been generated. GBS has had only a limited influence on policy
development processes at this level. Four particular factors would appear to be at play:
Firstly, the underlying capacity constraints from which Mozambique suffers –
noted in Chapter 2 as a key aspect of the context for Budget Support – mean that
policy analysis and development skills will always be at a premium. Capacity-
building inputs provided through the Budget Support process can help to address
these constraints (and we argue below that they have succeeded in reducing such
constraints in the PFM area) but there are limitations as to how far external agents
can help to develop internal capacity. Moreover, where Budget Support dialogue is
too wide-ranging, it may inadvertently dissipate the limited existing capacity for
policy analysis and development.
Secondly, sector level processes – as we have seen with the case studies of
agriculture, education and health – have their own history and their own dynamics,
on which the influence of the central level GBS PAF is modest at best. Where the
sector dynamics are good, such as in the case of education, the GBS dialogue can
help to sharpen the strategic focus and to raise awareness of sector level issues;
where the sector dynamics are less favourable, such as in agriculture, a more
interventionist approach would be needed in order to have any significant effect:
this may not always be possible and does not carry any guarantee of success.
Thirdly, the disinvestment in the Budget Support process, reported by most
stakeholders to have afflicted both the DPs and Government, has clearly reduced
the administrative attention and analytical effort which might potentially be
harnessed to address sector level policy problems. For Government, the cause
would appear to be the broadening of the Government’s own agenda, particularly
due to the need to address the policy requirements of the minerals and energy
sector. For DPs, this would appear to be a consequence of declining administrative
budgets and HQ pressures to reduce the degree of concentration on Budget
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 76
Support. Whatever the cause, this is a phenomenon to which most stakeholders
attest, and which few believe will be reversed. In these circumstances, policy gaps
at the sector level can only be effectively addressed through careful prioritisation
of the areas where policy analysis should be focused and/ or through a more
careful division of labour – both within government and within the G-19 – so that
available human and administrative resources are more efficiently used.
Finally, many DP stakeholders have asserted that there has been a decline in the
quality of coordination within the G-1934. It is said that there has been a
consequent decline in the ability of the G-19 to develop well researched and
coherently defined policy positions and to harness analytical and financial
resources for the common good. Unfortunately, the evaluation team have been
unable to determine the root cause of these problems from the evidence available
but it seems clear that a concerted effort to revitalise coordination within the G-19
will be needed if more effective attention is to be given to the policy gaps at the
sectoral level.
3.3. Contributions to capacity building
131. The provision of technical assistance and capacity building is generally perceived as
the third element of the “Budget Support package”. Such inputs would normally be focused
on aspects of policy, institutional reform and monitoring linked directly to the Budget
Support dialogue - such as support to the monitoring of the PRSP and to the development
of statistics, or to PFM reform.
132. In Mozambique, there is only a small minority of Budget Support operations which
make direct provision for the supply of technical assistance and capacity building. From
2011 onwards, only KfW provided such support in this way, while, between 2005 and 2010,
only 2 or 3 per year of the 17-19 DPs providing Budget Support made provision for
technical assistance and capacity building as part of their Budget Support financing
agreements. In these cases, they were general provisions, which could also be used to
finance studies (such as PEFA assessments), as well as technical assistance to support
monitoring and evaluation tasks undertaken for the Budget Support provider. The three
agencies, who were providing TA support in this way in this period – Danida, the EU and
KfW, did not provide detailed expenditure data on how these facilities had been utilised.
133. Some agencies have funding lines for Technical Assistance, which are managed
separately from Budget Support financing agreements but used to provide policy analysis,
34 Statements to this effect were made by several well informed DP stakeholders both during interviews
undertaken in the field mission and in the comments provided on the first draft of this report.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 77
advice and specific technical assistance in GBS relevant areas. For example, DFID report
that such a funding line has been used to fund technical assistance to the Public-Private-
Partnerships unit in MPD, to provide support to the drafting of legislation on State-Owned
Enterprises, to help establish the EITI Secretariat and to support the Ministry of Justice in
its anti-corruption work. Other G-19 agencies have similar budget lines of this kind for
GBS-related technical assistance but the evaluation team were unable to obtain
comprehensive documentation on such initiatives.
134. More structured programmes of technical assistance and capacity building have been
provided through common basket funding arrangements established to help create
capacity in key sectors or institutions. These were essentially of three types:
Support to PFM and governance functions through the common funds established
for (i) UTRAFE/ CEDSIF to cover the requirements of the integrated financial
management system (e-SISTAFE) and the PFM reform programme; (ii) the
Autoridade Tributária de Moçambique (ATM – the Tax Authority) to support revenue
administration reforms; (iii) the Tribunal Administrativo to support the
development of external audit capacity; and (iv) the Inspeção Geral de Finanças to
support the development of internal audit capacity.
Support to the national statistics system common fund, focused on the
development of statistical capabilities, linked to the monitoring and evaluation of
the poverty reduction strategy (PARPA I, PARPA II and the PARP). This common fund
was primarily focused on support to the National Statistics Institute (INE) but has
also provided funding for research studies undertaken by other organisations.
Support to the main sectors, notably Agriculture, Education, Health and Roads to
develop sector policy and planning capabilities and monitoring and evaluation
capacity, as well as, in some cases, financial and human resource management
capabilities.
135. Of these three types of arrangements, the third type – linked to sector capacity
building – never had any significant links to Budget Support and grew in most cases out of
well-established SWAP arrangements in these sectors, such as FASE, PROSAUDE and
PROAGRI.
136. The common funds linked to PFM/ Governance functions and to statistical capabilities
had their genesis in the early period of Budget Support between 2000 and 2004 (Batley, R.
et al, 2006). As a funding arrangement linked directly to PARPA I, many of the targets in
the first PAFs for Budget Support related to improvements in these areas. At the same
time, the deficiencies in capabilities were abundantly clear and accordingly the Government
sought TA/ capacity-building support, which was provided through these common funds.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 78
137. By the beginning of our evaluation period in 2005, these funds had taken on an
institutional identity and a momentum of their own. Indeed, when interviewed, Government
and DP stakeholders stated that they believed these funds would continue to exist, even if
Budget Support were to stop. The link to the Budget Support process is thus less strong35
than it was but these TA funds clearly operate as a complement to Budget Support, and we
therefore judge that it is still appropriate to conceive of these arrangements as Budget
Support inputs.
138. Taken together, the inputs provided by these common funds, alongside the other
funding lines linked more directly to Budget Support operations, amount to a significant
TA/ capacity-building input within the Budget Support package. Unfortunately, due to the
multitude of these arrangements and the range of funding lines through which they have
been provided, we have been unable to receive the detailed project documentation,
including progress reports and spending data, which would be necessary to make a
judgement about the efficiency and effectiveness of these TA/ Capacity-building inputs.
However, Government stakeholders were all positive in their appreciation of these inputs,
both with regard to their relevance and the relative efficiency with which they were
provided.
139. There is evidence that the TA/ Capacity-building provided by the common funds did
generate important outputs, which in some cases facilitated the production of the “induced
outputs” targeted by Budget Support. This has been most notable with regard to PFM
reforms and the strengthening of the capacities of the Autoridade Tributária de
Moçambique (ATM), the Tribunal Administrativo and the Inspeção Geral de Finanças. We
analyse these “induced outputs” in more detail in Chapter 4.
3.4. Adapting the Budget Support design to the changing context
140. Evaluation Question 1.2 asks how the context for Budget Support has evolved over the
evaluation period and whether the changes made in the scale and in the design and
implementation arrangements for Budget Support have been relevant to the evolving
context? In order to address this question, we first present a recapitulation of the main
contextual changes over the period. We then identify the key changes introduced in the
design and management arrangements. Finally, we consider the consequences of these
adaptations and the extent to which they represented an appropriate and adequate
response to the changed context.
35 For example, within our evaluation period only KfW provided support to these common funds through
the Budget Support agreement itself. Funding by all other agencies was provided through separate financing agreements.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 79
Changes in the context for Budget Support over the evaluation period
141. As detailed in Chapter 2, the principal contextual changes over the period 2005 – 2012
may be summarised as follows:
The impact of the global financial crisis on Mozambique was relatively limited.
There was a fall in economic growth during 2008-2009 but Mozambique’s rate of
economic growth bounced back quickly, due to a countercyclical policy stance,
increased donor support, particularly through the IMF’s Exogenous Shocks Facility
(ESF) and an increase in lending by the domestic banking system. By 2010, export
levels had recovered and growth rates had returned to previous levels.
The main economic transformation in Mozambique has been the discovery over
2010 – 2012 of very large deposits of exploitable gas in the Rovuma Basin. The
planned investments in the Rovuma basin, combined with the requirements for the
increased exploitation of gas reserves in the Mozambique Basin, and of coal and
heavy sands, amount to total FDI of some US $88 billion over the next ten years,
more than six times the current GDP. Significant domestic revenue flows from
resource exploitation are unlikely to occur before 2020 but this “mining boom” has
the potential to raise per capita income levels dramatically and to provide a major
new source of domestic revenue.
Although the mining boom remains in the future, by 2013 steady growth in GDP
and in domestic revenue capacity had left Mozambique with an unprecedented
level of fiscal space. Economic growth averaged 7.3 % per annum throughout the
evaluation period, and the revenue to GDP ratio increased from 15.3 % in 2006 to
23.3 % in 2012 (IMF, 2014). Moreover, in 2013, the government received
substantial windfall revenues from the taxation of capital gains on the re-sale of
exploration contracts within the Rovuma Basin (equivalent to 4.2 % of GDP),
creating a significant capital investment fund.
The evaluation period has seen a growing concentration of political power,
combined with more serious electoral irregularities, and more widespread incidents
of harassment of civil society. The ruling party, Frelimo increased its share of
parliamentary seats from 129 out of 150 in 1994 to 191 in 2009. Voter turn-out
halved over the last four elections, meaning that the substantial majority of 2009
was obtained with less than 30 % of the registered voters. There is evidence that
this concentration of power has distanced government from citizens and reduced
the need to listen to criticism and to alternative policy voices. The irregularities
noted in the 2009 elections, where observer groups were highly critical of many
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 80
pre-election and election day processes, provided a tangible manifestation of this
change. The campaign period leading up to the 2009 elections was also marred by
a significant level of partisan violence, as were the municipal elections of 2013.
Some commentators (CESC, 2013) suggest that harassment and intimidation of
civil society has increased, a view supported by a number of survey-based indices.
The relative shift in European opinion at public and official levels regarding the
desirability of Budget Support has also represented an important change in the
context for Budget Support. This is evidenced in the international withdrawal from
Budget Support by the Netherlands, and in the changes in Budget Support policies
introduced by the European Commission, Sida, SECO and DFID. These changes have
promoted more rigorous assessments of fiduciary risks and attempts to develop
more explicit links between disbursement modalities and the results of Budget
Support.
Responses to the evolving context
142. Four types of changes have been introduced in the design and implementation
arrangements for Budget Support, which we examine in turn below:
A reduction in the scale of Budget Support disbursements relative to GDP and
relative to total public expenditure;
Increased use of variable or performance tranches, as a disbursement mechanism
for Budget Support;
Changes in the focus of dialogue, bringing new strategic issues into the Budget
Support agenda;
Increased attention to the underlying principles for Budget Support, and an
increased readiness to suspend disbursements when a risk of violation of
underlying principles is perceived.
143. In terms of scale, over 2008 to 2012 disbursements of Budget Support remained
broadly stable at an average of $464 million in nominal US dollar terms, having risen by
75 % from 2005 to 2008. (Table 7) Although this represents a modest decline in real
terms, the relative stability of disbursements suggests that the falling support for Budget
Support noted amongst the majority of the bi-lateral agencies has not impacted
significantly on the magnitude of flows to Mozambique. The relative importance of Budget
Support has, however, declined in recent years both as a proportion of GDP, from 5 % in
2007 to 3.1% in 2012, and more significantly as a proportion of public spending (18.1% in
2007 to 9.4% in 2012; Table 8). However, this change has been driven by the fast growth
in domestic revenues and in public expenditures.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 81
144. One aspect of the changing design of Budget Support has been the increased use of
“variable tranches” (VTs) or “performance tranches”, in which the value of disbursements is
linked either to the completion of certain reforms or to the achievement of specific service
delivery outcomes. The EU, Sweden and Switzerland employed such mechanisms from the
outset of the period but Denmark, Germany and most recently the United Kingdom have
incorporated them during the evaluation period36. The relative importance of these variable
tranches has remained quite modest37. The actual size of the variable portion for the six
Development Agencies using the instrument is shown in Table 14. In 2012, the maximum
variable tranche disbursement would have comprised $70 million, of which $49 million
were paid – approximately 11 % of total Budget Support disbursements.
Table 14: Variable tranche disbursements, 2012 - 2014
Source: Orlowski, 2013
145. As can be seen from Table 14, the rate of disbursements of variable tranches was 70
% in 2012 and 69% in 2013, reflecting the fact that the majority of indicators were in fact
attained. Can we conclude from this that the increased use of variable tranches had a
positive incentive effect on government reform efforts? It has been impossible to
undertake a comprehensive assessment of this question essentially because it is too
complex a task to undertake with limited resources. There have been regular changes in
the definition of indicators across the evaluation period, the range of policy issues involved
is broad and the web of underlying causality mechanisms with which VT-induced incentives
would need to interact is too complex to untangle easily. However, we can point to two
sources of evidence:
36 Interestingly, Ireland abandoned its use of variable tranches during the evaluation period. We have
unfortunately not been able to obtain documentation on the reasons for this change.
37 We are indebted to Dieter Orlowski for the provision of information presented in this section: Orlowski, D. (September, 2013), Impact of variable tranches on policy and reforms, Internal Note for Danida, Maputo.
2012 2013 2014
potential granted potential granted potential granted
Donor's currency (million)
Denmark DKK 22 18 22 13.2 22 16.5
UK GBP no variable tranche 7 4 7 4.4
Germany EUR 10 8 5 4 5 not known
Switzerland CHF ? ? ? ? 4 3.2
Sweden SEK 180 115 90 67.5 no contract
EU EUR 21 14.375 21 14.375 21 14.375
USD (million)
Denmark 3.8 3.1 3.9 2.3 3.9 2.9
UK no variable tranche 10.8 6.2 10.8 6.8
Germany 12.9 10.3 6.6 5.3 6.6 not known
Switzerland 4.3 3.4
Sweden 26.6 17.0 13.8 10.4 no contract
EU 27.0 18.5 27.6 18.9 27.6 18.9
All VTs in USD 70.2 48.8 62.7 43.0 53.1 32.0
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 82
The data as presented in Figure 14 in section 3.2 do not suggest any trend
improvement in performance against PAF indicators since 2010, when variable
tranches have been more significant. From 2006 to 2009, 45-60% of indicators
were assessed as achieved, and only 10-15 % as not achieved, with a relatively
large group (25-40%) assessed as ‘not achieved but with some progress’. Over
2010 – 2012, the number assessed as achieved remained broadly the same at 45-
55% but the number assessed as not achieved rose sharply approaching 30 % in
2012, with a corresponding decline in those assessed as ‘not achieved but with
progress’. Clearly, there are several factors at play, including potentially more
challenging targets and/ or more rigorous standards of assessment, but the simple
numbers do not suggest a trend improvement in performance.
A more revealing body of evidence emerges from an internal study done for
Danida to assess the potential incentive effects of variable tranches. (Orlowski,
2013) Interviews were undertaken with the DPs using variable tranches within
their Budget Support operations as well as with officials from MPD and the
Ministry of Finance. As may be seen from Box 2, the results do not suggest any
systematic incentive effects.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 83
Box 2: Stakeholder’s feedback on the Incentive effects of Variable Tranches
Have variable tranches had positive incentive effects in Mozambique?
About two thirds of the variable tranches have been committed and subsequently disbursed in
recent years. Does this suggest that incentives have worked? We wanted to know from donor
representatives whether and how the variable tranche had contributed to this success.
Surprisingly, no one had a convincing story to offer. Rather, as one interviewee put it, a
“happy confluence of events” resulted in difficult indicator targets being attained. One
example is the internal revenue target. Most donors use this indicator as a trigger for the
variable tranche. Yet, several motives drive the revenue collection effort: notably, there is a
strong political interest in increasing public spending and reducing aid dependence and the
target is an essential parameter in the IMF programme. As a result, the target is usually met,
but probably not because of the additional incentive of receiving variable tranches.
Anti-corruption and procurement triggers appear in most donors’ variable tranches:
progress is visible although still below expectations. Was progress due to the incentive
effect of a variable tranche? Interviewees indicated that progress was made because of the
threat that the donor appetite for GBS would fade quickly. Changes were prompted by the
wider threat, not by the immediate effect of non-payment of variable tranches.
There was anecdotal evidence of effects on internal decision-making processes. The
Ministry of Planning called a donor in order to clarify the exact definition of the target so
that it would not be missed for lack of precision. A line ministry was called by the Ministry
of Finance, to ensure that the sector would meet a target linked to a variable tranche.
We had expected the existence of an internal version of the PAF indicator table with
amounts shown in the line of each indicator: it does not appear to exist. The PAF table is
apparently monitored regularly by the Council of Ministers, and indicators linked to
variable tranches are shaded; but the table does not show disbursement amounts.
We had also expected reports about the possible impact of incentives on decision-making
within sectors and were interested in whether variable tranches have in fact strengthened the
position of reform champions. No one had a story of this sort to tell.
Officials of the Ministries of Finance and of Planning stated that indicators to which
variable tranches are attached do not receive higher attention than others. The
government has defined its policies and uses a set of indicators to monitor their
implementation. Giving special attention to areas where financial rewards can be attained
would degrade others to second-class indicators, and should therefore be avoided.
Officials accept that donors have variable tranches essentially for domestic reasons,
and see this as an element of risk. They were opposed to the idea of donors bunching up
behind a reduced set of triggers, with higher values attached to each, because this could
reduce predictability. The potential variability was presented as an unavoidable risk, not as
a challenge to ensure predictability by influencing and closely monitoring results.
Source: Orlowski, 2013.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 84
146. Although indicative rather than conclusive, the evidence suggests that variable
tranches do not provide any significant additional incentive to meet reform targets. There
are already factors favouring reform progress – which is to be expected given that these
are targets agreed by Government and linked to national strategies and plans - but there is
no sense that those targets, to which DPs attach variable tranches are given any extra
priority. Indeed, Government stakeholders in interviews with Orlowski and with the
evaluation team quite pointedly stated that they considered it inappropriate to give higher
priority to some PAF targets over others.
147. Moreover, where Government has failed to meet targets linked to variable tranches,
there is no clear evidence that this has been due to a lack of political will or administrative
effort. Experience with the EU’s mechanism for determining the Millennium Development
Grant tranche is instructive in this respect. Under this mechanism, performance against
social sector outcome indicators over 2008–2010 determined disbursements for the years
2012–2014. The amount at stake (€15 million per year, €45 million in total) was
significant. The number of indicators (six) was limited, the mechanism was clearly defined,
and announced well in advance. The targets were drawn directly from sectoral and national
plans and they were neither controversial nor politically difficult. Yet, the assessment in
2011 was mixed: performance, as measured by the degree of achievement of targets
declined over the years and, as a result, Mozambique lost an annual €6.6 million of
potential budget support in each of the years 2012 – 2014.
Figure 16: Assessment of Performance to determine EU MDG tranche for 2012-2014
Source: Presentation by the EU on 02 June 2011, as quoted in Orlowski, 2013
8
3. MDG Tranche – Performance on MDG Tranche (1/2)
Subset of Indicators for the MDG Tranche 2008 2009 2010
Net enrolment at 6 years of age in Grade 1
- Girls 0,5 0,5 0,5
Conclusion rate in primary school, second
level (EP2) - Girls 1 0,5 0
Coverage rate of institutional births 1 0 1DPT3 and HepB coverage rate in children
between 0-12 months 0 0,5 0
Number of inhabitants per qualified health
worker 1 1 0,5
Total number of people (older than 15)
benefiting from Anti-Retro Viral Therapy 0,5 1 1
67% 58% 50%
20%
30%50%
Total Score for MDG Tranche = 55,83%
MDG Tranche = 55,83% * € 45m = € 25,125m
(or € 8,375m/ year instead of max of €15m)
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 85
148. Yet a closer examination of the targets reveals a set of relatively complex policy
problems. For example, there are a variety of reasons – social, economic and cultural - why
both matriculation and completion rates for girls tend to be low in Mozambique. Designing
specific public policy interventions to address this problem requires a detailed
understanding of its causes and of how those causes may vary between rural and urban
areas and by locality. Dedicating resources to research and to pilot testing of potential
policy responses might thus have proven more helpful than the provision of a variable
tranche incentive mechanism. This is especially probable in the light of the known
weaknesses in the policy research and development process within Mozambique: to expect
this process to generate new and more effective policy solutions simply as a result of a
variable tranche incentive seems somewhat unrealistic.
149. The increased use of variable tranches in Mozambique has been primarily driven by
policy requirements from DP headquarters to show a closer link between Budget support
disbursements and the achievement of results. In addition, the selection of indicators
seems to reflect concerns over corruption and governance. Only the EU still attaches
variable tranches to service delivery, in health and education: attention has shifted to
governance issues: in particular, to indicators in the areas of anti-corruption, transparency
in natural resource management, procurement, external audit and internal controls.
150. There is no clear evidence that the increased use of variable tranches has generated
incentives to give increased attention or to dedicate additional resources to the related
reforms and policy actions. Moreover, increased use of this mechanism has raised
transaction costs (albeit fairly modestly to date) and there is a risk that it may undermine
the predictability of Budget Support. In addition – and perhaps most significantly – it may
have served to divert attention from addressing underlying problems in the quality of
research and policy development processes at the sector level.
151. In managing the implications for Budget Support dialogue of the discovery of the
extensive gas reserves in the Rovuma Basin, a different approach has been adopted. The
design of the future fiscal and institutional framework for natural resource exploitation
has been taken up as an important aspect of the Budget Support dialogue. At one level, the
way in which this was done was quite “traditional” – a PAF indicator on progress to EITI
compliance was developed and indicators related to transparency of the fiscal
arrangements for natural resources were included in the variable tranches. But much wider
efforts were made to promote debate and thinking on the optimum design of
arrangements. For example, various papers on the future management of natural resource
revenues were presented and discussed at the Nordic- Mozambique conference on inclusive
growth of May 2012, study tours were organised to examine the experience of other
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 86
countries and substantial advisory support was provided by the IMF and by Norway.
152. Thus, the approach adopted by the G-19 was to raise the profile of the natural
resource management issue and to provide exposure to international experience, not to
promote any specific institutional solution. Partly as a consequence of this approach, in
October 2012, after 3 years as a candidate country, Mozambique was granted the status
of a fully compliant member of the Extractive Industries Transparency Initiative (EITI).
Without doubt, this was an important step towards the establishment of transparent
contractual and taxation arrangements.
153. The other substantial change in the management of Budget Support has been an
increased vigilance over the respect for the underlying principles of Budget Support,
culminating in three cases of partial suspensions of Budget Support disbursements. The
most notable – and long lasting of these - was provoked in particular by concerns of
irregularities in the conduct of the 2009 elections. Although the EU and other observer
groups reported that the process of voting was conducted in a peaceful and orderly
manner, they were critical of the pre-election processes. In particular, it was pointed out
that the National Electoral Commission (CNE) decision to disqualify, for ostensibly technical
reasons, the MDM candidates’ nomination papers in 9 of the 13 parliamentary
constituencies had substantially restricted voter choice38. Observers also documented
irregularities with election day processes, including ballot stuffing and tabulation fraud at
some polling stations, although these distortions were considered insufficient to have
affected the overall result of the election. (Freedom House, 2010).
154. The electoral irregularities reported, combined with on going concerns about slow
progress in the fight against corruption, prompted the G-19 Budget Support group to report
to Government that it felt that a breach of the underlying principles for Budget Support
might have occurred. This led to the suspension of Budget Support disbursements by 4 DPs
and an intense process of dialogue between Government and the G-19, which became
known as the period of “crispação” (“political tension”). As a result, Government agreed to
implement a Governance Action Plan, whose key elements included the preparation of
revisions to the electoral law (drawing on recommendations from the Constitutional
Council, election observer missions, the Head of State and other stakeholders), certain
measures to promote political inclusivity, and a package of Anti-Corruption legislation. In
addition, an informal agreement was reached for the Government to promote the
acceptance of the MDM as a formal opposition “bancada” (“bench”) – a proposal, which the
parliamentary authorities accepted. Normally, MDM would have required a minimum of 10
38 As a rough indication of the impact this might have had, if MDM had won the same percentage of
parliamentary as presidential votes (8.59 %), they would have won some 15-20 seats, as opposed to the 8, they actually won. Whether this would have been at the expense of Frelimo or Renamo is difficult to judge.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 87
seats to obtain the status of a bancada; without such a status, their rights to debating time
and other such parliamentary privileges would have been significantly restricted.
155. There were two further suspensions of disbursements due to concerns over underlying
principles, prompted by health procurement irregularities in 2011 and by education payroll
irregularities in 2012. There is some controversy over whether these three suspensions
were justified. The Budget Support MoU of 200939 includes a commitment by DPs to “not
interrupt in-year disbursements unless there is a violation of the underlying principles”. In
each of these cases, there were grounds to suspect that a violation of the underlying
principles might occur but subsequent corrective actions by Government prevented a
formal violation of underlying principles from being declared. Government stakeholders
maintain that the “pre-emptive” suspension of Budget Support by certain partners was not
needed, while several DPs expressed a conviction that corrective actions would not have
been taken or would have been delayed in the absence of the decision to suspend
disbursements.
156. We would judge that the suspension of payments in early 2010 was instrumental in
generating corrective actions at the time of the “crispação” but not necessary at the time
of the revelation of the education payroll fraud in late 2011/ early 201240. Problems with
the education payroll were identified by the Ministry of Education itself and both corrective
actions and further investigation of the source of the problems had been initiated by the
Government before the G-19 became aware of the problem. Under these circumstances,
there would not appear to be grounds for suspension of Budget Support, even if there were
good grounds for a discussion over the appropriate corrective measures and for a process
of follow-up to these measures – through the adoption of payroll reform as an ‘area for
special attention’.
157. The financial costs of these temporary suspensions were not so great but there may
have been more serious and potentially lasting negative effects on the quality of the
Budget Support partnership. The Budget Support partnership requires that disbursements
should only be suspended in extreme circumstances, because the guarantee of a
predictable flow of resources is fundamental to the partnership. Moreover, Budget Support
supports the ongoing strengthening of PFM systems and of anti-corruption mechanisms: as
these systems become stronger, the level of detection of fraud and mismanagement is
likely to improve; to suspend Budget Support each time such a case is detected would be
akin to “shooting the messenger” for bringing bad news. What is important to the Budget
Support partnership is that cases of fraud should be made public and that adequate
39 Point 25 of the Memorandum of Understanding (2009).
40 Information on the circumstances of the 2011 suspension was not sufficiently clear to form a judgement.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 88
corrective actions should be introduced and vigorously pursued. The absence of such a
process might legitimately be considered a violation of underlying principles but the
occurrence of fraud should not, in itself, be considered as such.
Conclusions on the appropriateness of responses
158. To summarise, our judgements on the appropriateness of the corrective actions
introduced in response to the changing context for Budget Support are as follows:
It seems appropriate that the relative scale of Budget Support has been allowed to
diminish as domestic revenues have grown. Not only has this ensured that
investments and service delivery processes previously financed by Budget Support
can now be sustained with domestic financing, it has also helped to reduce the
unusually high level of aid dependence, which had previously prevailed. However,
the growth of domestic revenues has now generated a new opportunity for
Mozambique by creating the fiscal space necessary to undertake more ambitious
public spending actions to reduce poverty. If the Government of Mozambique can
demonstrate its ability to allocate and implement an expanded public budget
effectively, then there is every reason to maintain or even expand the current
scale of Budget Support, given that such spending should be sustained by a
growing domestic revenues over the medium term.
There is no evidence that the expanded use of variable/ performance tranches has
generated positive incentive effects for the implementation of reforms or the
achievement of specific service delivery results. It has allowed some Budget
Support providers to signal to their domestic constituencies the nature of the
results, which Budget Support is expected to generate. In the process, it may also
have helped to improve the understanding of the Budget Support mechanism. If so,
it would probably have been a worthwhile response to domestic concerns over the
effectiveness of Budget Support. Moreover, within the period of the evaluation, the
increased transaction costs, which these variable tranches have entailed, do not
appear excessive and the relative impact on the predictability of Budget Support
also appears limited. However, there is no evidence that these variable tranches
have generated positive incentive effects for the implementation of reforms.
Moreover, there is a danger that the growing interest in the use of such
mechanisms may detract from efforts to provide direct support and advice to the
government in the formulation and implementation of public policy. There continue
to be significant weaknesses in the capacity to design and implement policy
reforms and it is important to ensure that Budget Support dialogue, and
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 89
accompanying capacity building efforts, are focused on resolving these
weaknesses as effectively as possible.
The support provided to the Government to address the legal and institutional
requirements for managing natural resource extraction and the accompanying
revenues is a good example of the creative use of GBS dialogue and accompanying
technical assistance. The G-19, along with the IMF, were quick to identify this need
and provided support in an appropriate manner. It seems unlikely that progress
towards EITI compliance would have been as swift as it was in the absence of this
support.
There are indications that the apparent readiness to announce GBS suspensions on
the grounds of potential violations of underlying principles is undermining the trust
between Government and the G-19, and the quality of the related dialogue. The
evaluation period saw three cases of partial and temporary suspension of Budget
Support disbursements on the grounds that there was risk of underlying principles
being violated. In the first of these cases – the ‘crispação’ of late 2009/ early
2010, we would judge that the suspension was justified and effective in
generating corrective actions by government, which would probably not have all
been undertaken in the absence of this suspension. In the latter two cases, in 2011
(prompted by irregularities in health procurement) and in 2012 (prompted by
education payroll irregularities), the justification for suspension is less clear, and in
2012 in particular, does not seem to have been justified. The immediate financial
implications of these suspensions were not so significant but the effects on the
quality of dialogue may have been more serious. A measured response to cases
where fraud or corruption is detected is essential, if open communication and
constructive dialogue is to be maintained. In this respect, the revision of the
approach to underlying principles, through the process of updating the GBS MoU,
seems a welcome step in the right direction.
4. Facilitating improvements in Government policies: Induced Outputs of Budget Support
159. The analysis of Chapter Three has provided the basis for understanding how Budget
Support has been inserted into the public spending and policy-making process and with
what immediate effects. This chapter completes Step One of the analysis, considering how
far Budget Support has facilitated improvements in Government policies and practices. It
thus considers the “induced outputs” of Budget Support, comprising level 3 of the
framework.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 90
160. Specifically, Chapter 4 examines the changes evidenced in the financing and
institutional framework for public policy, public spending and service delivery and assesses
how far these might be attributed to Budget Support. It considers: (i) the changes in the
overall macroeconomic and fiscal position of Government; (ii) the evolving status of PFM
systems; (iii) the evolution in the sectoral composition of public spending; and (iv) the
changes in sectoral policies in agriculture, education and health; and v) changes in the
institutions of democratic accountability and in the participation of citizens and
Parliamentarians in the budgetary process.
4.1. Macroeconomic management & revenue generation
161. Evaluation Question (EQ) 3.1 first examines the extent to which there have been
improvements in the quality of domestic revenue mobilisation and macroeconomic
management. Secondly, it considers how far such improvements can be attributed to
Budget Support and, if so, to which mechanism of influence. It also examines whether
Budget Support may have had any negative effects on macroeconomic management,
notably through “Dutch Disease” effects, or through problems of monetary sterilisation.
Domestic revenue mobilisation and macroeconomic management
162. Domestic revenues have increased rapidly throughout the evaluation period, as a
result of high levels of economic growth and reforms in tax administration. The revenue to
GDP ratio has increased significantly in the past 7 years from 13.4 per cent in 2005 to 23.3
per cent in 2012 (Table 16). At the same time, there has been a shift in the composition of
revenues: in 2007, taxes on goods and services comprised the majority of revenues
collected, but taxes on incomes, profits and capital gains have recently been increasing. In
2012, these outweighed revenue from goods and services taxes. Tax reforms have
strengthened the collection of income, profits and capital gains tax, while rising import
volumes have boosted import tax revenues and domestic sales taxes. (Figure 17.)
163. Domestic revenues have however been inflated by VAT figures which have been
calculated net of refunds. As a result, fiscal data over-estimate the levels of VAT receipts.
There is also a backlog of VAT refund requests, which equalled 1.3 per cent of GDP in 2013.
Moreover, the tax burden is quite narrow and unequally distributed with a heavier burden
on Small and Medium Enterprises (SMEs), while there are exemptions for megaprojects.
These concerns have generated discussion between the IMF and the Government over the
possibility of suspending the 0.5% target for annual revenue growth, and introducing
policies to ease the tax burden on small businesses.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 91
Figure 17: Revenue Collections 2006-2012 (% of GDP)
Source: MoF and IMF
164. In 2012 and 2013, Mozambique received significant windfall receipts from taxes on
capital gains arising from the re-sale by initial investors of exploration rights and shares in
the gas sector. In 2012 these were equivalent to approximately 1¼ per cent of GDP (IMF
2013 Jan) and, over the course of 2013, they amounted in total to some US$ 400 million,
about 2.5 per cent of GDP. How these windfalls will be spent is a matter of some concern,
because significant expansion in recurrent spending will not be sustainable in the medium
term, as further windfalls are not guaranteed. Discussions with the IMF are on going over
the design of mechanisms to ensure these resources are utilised predominantly for
productive investments.
165. Mozambique’s revenue performance compares well with other countries in the region.
Domestic revenue collection as a percentage of GDP is significantly higher than in Tanzania
(17.6 % of GDP in 2011/ 2012) and Uganda (15.6% in 2011/ 12) and compares favourably
with Kenya (23.5 % of GDP in 2011/12.) This represents an exceptional performance by
Mozambique, as Kenya has had a long history of tax reform and the size of its formal
sector is larger than Mozambique’s (Various IMF Article IV Reports).
0,00%
2,00%
4,00%
6,00%
8,00%
10,00% Taxes on income andprofits
Taxes on goods andservices
Taxes on internationaltrade
Other taxes
NONTAX REVENUE
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 92
Table 15: Minerals Sector Fiscal Frameworks for Mozambique & competing countries
Royalty Rates
Corporate Income Tax
Accelerated Depreciationi
Effective Tax Rateii
Mozambique 3-10% 32% 25% 52%
Botswana 3-10% 25% 100% 50%
Ghana 3-12% 25% 75% 54%
Namibia 5-10% 34% 33% 48%
Tanzania 0-5% 30% 100% 45%
Zambia 6% 25% 100% 47%
Source: Hubert, 2012
166. The mineral sector has the potential to contribute substantially to future revenue
growth, although major revenues are not expected for another decade. An important
element of this is to have a good fiscal regime for the mineral sector, to ensure revenues
are maximised. Work has already been undertaken on this with the existing regime
tightened. In 2013 the Government approved draft revisions to the mining and hydrocarbon
laws. As Table 15 indicates, Mozambique’s tax regime for the mineral sector has become
closer to international norms and is similar to the core provisions in other African countries
(Hubert 2012). However, despite contracts since 2007 needing to adhere to this new fiscal
framework there have been concerns that some projects are given “one-off” exemptions.
167. Domestic revenue from coal and gas will grow gradually, only exceeding current aid
flows of $1.2 billion in a decade’s time (IMF 2013). Over 2007-2011, investments in coal
and liquid natural gas (LNG) contributed an annual average of less than ½ percentage
point of GDP to total fiscal revenues, although coal royalty revenues have been more
significant in 2012 and 2013. This is illustrated in Figure 18 below, which shows the
projected levels of exports of these commodities (right-hand scale) and the coal and LNG
contribution to fiscal revenues (left-hand scale) as a percentage of non-coal, non-LNG GDP.
Over the short to medium term government revenue from coal and gas is projected to be
small. It is estimated that total revenue from coal will not exceed 2 percent of GDP,
whereas the LNG sector will only start production in 2020 and will not reach full capacity
until around 2023.
Figure 18: Mozambique : Coal and LNG Contribution to Fiscal Revenue, 2011-2032
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 93
Source: IMF 2013
168. Mozambique became a candidate for the Extractive Industry Transparency Initiative
(EITI) in 2009 and became compliant in 2012. This was an important step to allow for the
transparency of contracts, regulation, taxation and revenue (EITI 2012). This has resulted in
coverage of its reports on extractive industry payments and government receipts
expanding from 6 to 46 enterprises by 2013. Payments reported by mining companies now
correspond very closely to receipts recorded by government. In terms of EITI membership,
Mozambique is at a similar stage to Tanzania who also achieved EITI membership in 2012.
Table 16: Overview of Central Government Fiscal Operations 2005-2012 (% of GDP)
(Figures in % of GDP) 2005 2006 2007 2008 2009 2010 2011 2012*
TOTAL REVENUE 13.44% 14.98% 15.92% 15.92% 17.81% 20.15% 22.07% 23.25%
TAX REVENUE 11.02% 12.93% 14.12% 15.43% 15.61% 17.96% 18.69% 19.82%
NONTAX REVENUE 2.41% 2.04% 1.81% 0.49% 2.19% 2.19% 3.38% 3.42%
TOTAL EXPENDITURE 23.60% 26.00% 27.53% 27.73% 32.52% 31.68% 36.01% 32.63%
CURRENT EXPENDITURE 13.42% 14.24% 15.33% 15.66% 17.44% 18.87% 19.45% 19.19%
Compensation to employees 6.83% 7.21% 7.75% 8.02% 8.87% 9.24% 9.76% 10.18%
Goods and Services 2.90% 3.03% 3.16% 3.34% 3.40% 3.23% 2.96% 3.71%
Interest on public debt 0.82% 0.77% 0.61% 0.52% 0.51% 0.84% 0.98% 1.01%
Transfer payments 2.46% 2.53% 2.78% 2.91% 2.98% 3.00% 3.07% 1.99%
Other 0.41% 0.70% 1.02% 0.86% 1.68% 2.55% 2.67% 2.30%
CAPITAL EXPENDITURE 8.64% 10.34% 11.22% 11.54% 13.32% 13.14% 15.53% 12.34%
of which: locally financed 3.22% 3.64% 4.41% 4.55% 4.73% 6.06% 6.50% 6.06%
of which: ext financed 5.42% 6.70% 6.82% 6.99% 8.59% 7.08% 9.04% 6.28%
Net lending 1.43% 0.91% 0.79% 0.79% 1.66% 0.61% 1.00% 1.10%
Unallocated revenue(+)/expenditure(-) 0.11% 0.52% 0.18% -0.26% 0.09% -0.94% 0.03% -0.12%
Overall balance before grants -10.17% -11.03% -11.60% -11.81% -14.76% -11.59% -14.04% -9.50%
Grants received 6.55% 9.96% 9.05% 9.42% 9.50% 8.31% 7.84% 5.38%
Project 1.26% 2.42% 3.09% 2.56% 2.69% 2.03% 1.83% 1.79%
Non-project 5.29% 7.54% 5.96% 6.85% 6.81% 6.29% 6.00% 3.59%
Overall balance after grants -2.43% -1.06% -2.56% -2.39% -5.26% -3.28% -6.20% -4.12%
Net external borrowing 2.96% 3.92% 2.39% 2.88% 5.07% 3.82% 4.67% 3.39%
Disbursements 3.44% 4.52% 2.78% 3.19% 5.33% 4.15% 4.97% 3.79%
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 94
Project 2.07% 2.82% 1.54% 1.48% 2.45% 2.18% 3.85% 1.65%
Non-project 1.37% 1.70% 1.24% 1.71% 2.89% 1.97% 1.12% 2.14%
Cash amortization -0.48% -0.60% -0.39% -0.31% -0.26% -0.33% -0.13% -0.39%
Net domestic financing -0.22% -3.16% 0.15% -0.83% 0.29% -0.54% 1.50% 0.72%
Source: Bank of Mozambique, except for 2012 (IMF estimates from IMF, 2014).
169. Total expenditure has risen by more than 9 percentage points of GDP between 2005
and 2012, which is similar to the growth in revenue achieved. (Table 16.) Recurrent
expenditure has risen from 13.4 per cent of GDP in 2005 to an estimated 19.9 per cent in
2012. However, it has not outstripped revenue growth, hence, current levels of expenditure
should be sustainable.
170. Development expenditure increased from 8.6 per cent of GDP in 2005 to 12.3 per cent
in 2012, based upon an increasing level of domestically financed Development spending,
which rose from 3.2 per cent of GDP in 2005 to an estimated 6.1 per cent in 2012.
171. The overall balance before grants increased sharply in 2009 as a result of the global
financial crisis, but fell significantly in 2012, towards a more sustainable level. However, a
high proportion of deficit funding used to come from budget support grants and loans, and
from project grants and loans, both of which are decreasing. Net aid flows have declined
from 14.8 percent of GDP in 2009 to 12.8 percent in 2011.
172. External borrowing is likely over time to replace budget and project support for the
financing of capital spending, as investment needs will remain large. In the short to
medium-term government finances should benefit from on-going revenue reforms and in
the medium to long term from increasing revenue from investments in the extractive
industry.
173. Although Government should be able to sustain a higher level of borrowing in the
future, there is a risk of creating an imbalance between debt obligations and revenue
flows. The November 2013 dollar bond issue of US$850 million by a newly created
Mozambican state company EMATUM (National Tuna Company) gives some indication of
the potential risk. The interest rate on this loan was reported in the press to be 8.5%, as
compared with commercial loans, currently being obtained at 6.5% by Ghana – another
economy with expanding mineral resources. High interest rates of this kind, combined with
demanding principal repayment schedules may result in debt obligations being “front-
loaded” relative to revenue flows. There is also a need to plan public investment projects
carefully so as to ensure that they meet policy priorities and are executed efficiently. There
are on going discussions with the IMF over ways to improve the institutional arrangements
for public investment planning and management.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 95
174. Over the evaluation period, Mozambique has remained on-track with the IMF PSI
programme. The majority of the performance assessment criteria have been fully observed,
and structural benchmarks have been met, although occasionally with delays41.
Macroeconomic stability and the business climate
175. GDP growth remained high throughout the evaluation period, averaging 7.3 per cent
annual growth. As we have noted in Chapter 2, the Mozambican economy was not
particularly affected by the global financial crisis.
176. Year on year inflation in Mozambique was 2.2 per cent in 2012 - the lowest in SADC.
This also compares favourably to the sub-Saharan average and to inflation rates in the
SSA frontier economies. (Figure 18.) Inflation averaged 10.2 per cent between 2004-2008
and then declined significantly in 2009, before rising sharply again to 12.7 per cent in
2010 and beginning to decline again in 2011. This was due to a restrictive fiscal stance by
the government and favourable trends in international energy and food prices. The CPI
inflation rate is expected to have risen to 6 per cent by the end of 2013 due to food
related price increases, although this remains in line with the government’s inflation target.
Figure 19: Consumer Price Index 2000-2012, compared with SSA and frontier economies
Angola, Ghana, Kenya, Mauritius, Mozambique, Senegal, Tanzania, Uganda, Zambia, Zimbabwe,
Source: IMF 2012
177. Mozambique has made efforts to improve the investment climate, through recent
reforms related to improving business licensing and registration for small businesses. Over
41 Recent delays relate to excess liquidity in late 2012 due to higher than programmed bank deposits at
the Central Bank, thus missing the assesment criterion on reserve money. Net international reserves also exceeded programme targets in 2012. The government revenue target was just missed in 2012, due to lower than expected inflation and problems with the introduction of the single-trade customs window (IMF 2013b).
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 96
the evaluation period, improvements were made in the time needed to start a business. The
number of days needed for this fell from 153 in 2005 to 13 in 2011. However, the number
of days taken to get electricity increased from 77 in 2010 to 107 in 2012, while there has
been little change in registering a property, getting credit or enforcing contracts.
Figure 20: Global Competitiveness, 2012-2013: Mozambique & neighbouring countries
178. Mozambique still ranks low on international indicators of economic competitiveness
and the business environment. In the Doing Business Index, Mozambique has remained at a
ranking of 138 or 139 out of 183 countries, throughout the evaluation period. In the World
Economic Forum’s Global Competitiveness Report 2012-13, Mozambique was ranked 138
out of 139 countries in terms of economic competitiveness, a fall from 133 in 2011-2012.
This compares unfavourably with neighbouring countries such as Uganda, Tanzania and
Kenya, who are all ranked higher than Mozambique. (Figure 20)
179. The Global Competitiveness Report 2013-2014 highlights the need for Mozambique to
invest in infrastructure in order to increase long-term competitiveness. It also notes the
importance of establishing a regulatory framework that encourages competition and
fosters economic diversification.. The Doing Business Indicators 2012 and 2013 point to
additional problems that constrain business activity, such as obtaining an electricity supply,
enforcing contracts, registering property and resolving insolvency. However, the 2014
Doing Business Report notes that improvements were made in 2013 to make it easier to
apply for and receive construction permits and licences for cross-border trading.
Contribution of Budget Support funds to macroeconomic management
180. During the evaluation period, there has been a strong performance by Mozambique in
macroeconomic management:
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 97
Total spending has increased by 9 points of GDP as a consequence of both
expanded recurrent and capital expenditure, while domestically financed
development expenditure has almost doubled from 3.2 % of GDP in 2005 to 6.1 %
in 2012.
Recurrent expenditure is estimated to be 19.2 per cent of GDP in 2012. This is a
level that could be sustained by domestic revenues alone.
Domestic revenue mobilisation has increased by 9.8 percentage points of GDP over
the evaluation period.
181. It is unlikely that the growth of domestically-financed capital expenditure could have
been achieved in the absence of budget support, without recourse to expensive and
probably unsustainable levels of borrowing. Until the discovery of gas in the Rovuma Basin
in 2011, Mozambique’s access to foreign commercial borrowing was quite limited.
Domestic borrowing was also constrained by the small size of the market for Treasury Bills,
and by the fact that in 2005 and 2006, the Government was carrying repayment
obligations due to the bank recapitalisation that took place in 2002. It might have been
possible to seek grant or concessional financing for the domestically financed capital
investments but foreign financed capital investments averaged 7.1 % of GDP through the
evaluation period. It seems implausible that there would have been the capacity to raise
foreign financing significantly above this level, given the complexity of the processes for
procuring such finance and the transaction costs and time lags involved.
Potential unintended negative effects of Budget Support funds
182. We have also examined whether these budget support flows had unintended negative
effects. There are three main ways in which this might have happened:
Through disincentive effects on domestic revenue collection;
Through “Dutch disease” effects, due to large foreign exchange inflows;
Through the inflationary effects of large monetary inflows or through the costs of
the monetary sterilisation measures taken to avoid these inflationary effects.
183. Following a brief gestation period in 2007 and 2008, when reforms to revenue
administration were being introduced, revenue growth has been consistent throughout the
evaluation period showing no clear relationship to fluctuations in Budget Support funding.
Budget Support funding may have a disincentive effect on domestic revenue collection
efforts by providing a source of budget funding, which does not carry the political costs of
increased taxation. Clearly, domestic revenue performance in Mozambique has been
impressive, but it is possible that in the absence of budget support, the incentives to
increase the mobilisation of domestic revenue may have been greater. However, the data
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 98
show a rate of revenue growth – particularly in tax revenues, which is probably without
precedent for a non-oil economy in Sub-Saharan Africa.
184. Discussions with the Authorities and with the IMF strongly suggest that the desire for
revenue growth has been unaffected by the presence of budget support as an alternative
funding source In discussions with the evaluation team, the Mozambican authorities
expressed a very strong commitment to reducing their reliance on external aid. As a result,
the government has made significant efforts to develop the tax regime, establishing a
clear fiscal framework for future taxation of extractive industries and renegotiating
exemptions for some mega-projects. These perceptions were confirmed by discussions with
the IMF, who have been consistent in their praise for the government’s efforts in revenue
collection. Indeed, during 2013, the IMF has encouraged the authorities to contemplate the
introduction of measures to reduce the tax burden, and has encountered some resistance
to this proposal.
Table 17: Real effective exchange rate and Budget Support receipts (2005-2011): 2005 2006 2007 2008 2009 2010 2011
Real Effective Exchange Rate (%
change on previous year)*
-6.3 1.8 6.5 11.8 -6.5 -15.1 19.7
Budget Support receipts (MZN
million)
6,214 8,824 10,364 11,047 11,513 14,999 15,886
* Positive values represent an appreciation, negative values represent a depreciation.
185. Dutch disease effects occur when aid flows (or large-scale export earnings due to
mineral exports) sharply increase the availability of foreign exchange, leading to an
appreciation in the real exchange rate. This reduces the income of the traditional export
sector, forcing its contraction. There has been considerable volatility in the real effective
exchange rate in Mozambique: it appreciated from 2005 until early 2009. It then
depreciated until the end of 2010, probably as a consequence of the global financial crisis
and the related policy responses. Since 2011, the real effective exchange rate has
remained above its average level but the IMF consider that this is likely to be at least
partly due to the start up of coal exports in 2011 (IMF 2013).
186. There has been no systematic appreciation of the real exchange rate during the
evaluation period. Nor is there any clear relationship to Budget Support inflows. (See Table
17.) It remains to be seen whether the real appreciation experienced since 2011 will push
the real exchange rate above its (previous) long-term trend level. Even if this were the
case, the fact that Budget Support flows have actually been decreasing since 2011 would
suggest that this would be due to factors other than budget support.
187. A comparison of budget support disbursements with fluctuations in Treasury bill rates
shows no indication of problems with monetary sterilisation. This can occur due to the
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 99
lumpiness of Budget Support disbursements. However, comparing these with Treasury Bill
rates, we see no evidence of “spikes” in interest rates in the periods immediately following
large Budget Support disbursements. This suggests that there have been no significant
sterilisation requirements – a fact confirmed in discussions with the Bank of Mozambique
and the IMF.
188. The main negative effect of budget support has occurred as a result of disbursements
being deposited at the central bank late in the fiscal year, with often large amounts of
other funds being transferred at the same time. These higher-than-programmed bank
deposits at the central bank have led to the IMF PSI target on reserve money being missed
in both December 2012 and March 2013. However, this was a minor deviation of 3 per cent
of reserve money or Mt 1 billion and was not particularly significant. It is also less a
problem of monetary sterilisation due to budget support and more a result of a lack of
coordination between the Bank of Mozambique and the Ministry of Finance. If the Central
Bank and Ministry of Finance had both been aware of when to expect budget support
payments, plans could have been made to increase foreign exchange reserves so as to
neutralise the impact of these payments on the money supply. An action plan has now been
drawn up to strengthen coordination between the two institutions and to ensure that this
problem is addressed in the future.
Budget support policy dialogue on macroeconomic issues
189. There is little evidence that the PAF indicators and dialogue relating to strengthening
revenue collection have influenced Government efforts in this area. Increasing revenue
collection was a major Government objective as well as an objective in the IMF Policy
Support Instrument (PSI) and the Government has had a good track record with the IMF
over many years. Budget Support has supported this objective, rather than being
instrumental in driving the momentum on this issue.
190. Similarly, interviews with the IMF, MoF and Bank of Mozambique confirmed the
government’s political commitment to macroeconomic stability. It therefore seems certain
that the Government would have pursued a stable macroeconomic policy even in the
absence of Budget Support.
191. The dialogue and indicators related to the private sector have supported government
actions in this area. Despite this, Mozambique has had a low ranking in global
competitiveness and business environment indices, with little improvement over the
evaluation period. We conclude that budget support dialogue has had little additional
influence in this area.
192. Stakeholders report that monitoring progress on the EITI initiative through the PAF
helped to add momentum to the process and thus contributed to the country achieving
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 100
compliance. We judge that Budget Support dialogue had a positive influence in this area.
There has been a significant amount of work undertaken by the G19 task force on
Extractive Industries, which was used for political dialogue with the Government. There was
also a PAF indicator from 2010, monitoring progress on the EITI initiative.
4.2. The quality of Public Financial Management
193. EQ 3.2 analyses to what extent there have been improvements in the quality of PFM
and how far these may be attributed to Budget Support. This question is addressed in four
parts:
Nature of PFM reform process and its relationship with the Budget Support
process;
Evidence of the impact of these reforms on PFM systems and processes, drawing
on the PEFA assessment findings (2006, 2009, 2011) and the UTRAFE / CEDSIF
annual reports;
Relevance of current reform programme, considering progress achieved to date
and challenges ahead; and,
Assessment of relative contribution to BS.
Relationship between PFM Reform and Budget Support
194. The Budget Support MoU sets out as an intermediate objective “providing financing to
the public sector for poverty reduction, clearly and transparently linked to performance, in
a way which improves aid effectiveness and country ownership of the development
process, reduces transaction costs, allows for allocative efficiency in public spending and
predictability of aid flows, increases the effectiveness of the state and public
administration, improves monitoring and evaluation and strengthens domestic
accountability”.42
195. This objective is in turn linked to the MoU underlying principles requiring from GoM the
establishment of sound PFM systems and their maintenance and improvement over time
(Section 4 of the MoU). 9 to 10 indicators have been regularly included in the PAF for PFM,
emphasising its high profile in the Budget Support dialogue.
196. The assessment of PFM performance against the PAF has in general been positive.
Over the 9 years from 2005 to 2013, 58 % of the PFM indicators were assessed as
achieved, 35 % as ‘not achieved but with progress verified’ and only 7 % as not achieved.
(Table 18). Interviews with GoM and DP stakeholders showed a high level of agreement on
42 GBS MoU, 2006, Section 2, Article 7 ii)
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 101
the relevance of the majority of the indicators, and in terms of the relative rankings of the
priority of each indicator, this was the area of the PAF, where DP stakeholders showed the
closest alignment with GoM and the greatest degree of harmonisation across DPs.
Table 18: Summary of PFM annual performance assessments in the Annual Reviews
Year Status per JAR Aide Memoire Status per PAF Matrix
2005 “Performance in this area is at best mixed.” 9 indicators out of which 2 achieved, 5 partially achieved and 2 not achieved
2006
“The overall assessment of GoM performance in the area of Public Finance Management is positive, although there remain some important areas of concern. The recently performed PFM Assessment shows that there are still important weaknesses in the PFM framework and addressing these will require continued efforts in the coming years.”
10 indicators out of which 8 achieved and 2 partially achieved
2007
“In global terms progress has been recorded in PFM in the course of 2006, but some areas remain a concern and are challenges to be considered in 2007 and subsequent years.”
12 indicators out of which 7 achieved and 5 with progress on going
2008 “The assessment shows that significant improvements were obtained in the PFM quality, during the 2004-06 period”
8 indicators out of which 5 achieved and 3 with progress verified
2009
“The performance of Public Finance Management (PFM) was positive in 2008. The Government is committed to continuing and deepening the development of PFM systems, and having ascertained the need for a long-term strategy, decided to prepare a Strategic Public Finance Management Vision, to be concluded during 2009.”
8 indicators out of which 5 achieved and 3 with progress verified
2010
“Overall there was progress in the area of Public Finance Management (PFM), essentially marked by the consolidation of reforms initiated earlier. Nevertheless significant weaknesses and challenges remain to be faced.”
8 indicators out of which 5 achieved and 3 with progress verified
2011
“Global performance was mixed. (…) The PEFA assessment carried out confirmed important progress in the period 2007 – 2009 but also highlighted the need to address important challenges related to the consolidation of reforms (…)”
9 indicators out of which 5 achieved, 2 with progress verified and 2 not achieved
2012 “In general, performance was considered to be satisfactory.”
9 indicators out of which 6 achieved, 2 with progress verified and 1 not achieved
2013 “A good performance, in general, has been observed in PFM as a consequence of many of the reforms under implementation (…)”
8 indicators out of which 4 achieved, 3 with progress and 1 not achieved
Evidence of Government commitment to PFM reforms
197. There has been a consistent government commitment to PFM reform since well before
the evaluation period. The development of the conceptual framework for the development
of the SISTAFE (System for Administration of State Finances) and the drafting of its
respective law and regulations represent tangible evidence of this, alongside the decision
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 102
to develop the e-SISTAFE IT tool as the principal vehicle for implementation of PFM
reforms. The key milestones within the PFM reform process are summarised in Table 19.
These form part of a wider set of institutional and technical reforms, effectively initiated in
2002.
Table 19: Key Events & Developments in PFM (including the IFMIS)
Date Key Event / Development in PFM (including IFMIS development from 2004 onwards)
2002 UTRAFE is established within MoF
GoM starts to submit annually its Plan (PES) and Budget (LOE) to Parliament
New PFM system introduced (Law 9-2002)43
2003 Public Expenditure Review by World Bank
2004
Decree 23-2004 approving SISTAFE Regulation
1st External Audit Report by TA (CGE 2002)
MoU for GBS signed and PAF adopted
Ministry of Finance and Plan separated into Ministries of Finance and Plan & Development)
Institutional development of UTRAFE and conclusion of first stage of e-SISTAFE development (November 2004)
2005
Introduction of new Procurement Regulations
Introduction of e-SISTAFE Rolling Plan (with gradual introduction from central to provincial and district levels) and move from advancement of funds to direct budget execution (process still on-going)
CUT in Meticais operationalized and the accounting module in e-SISTAFE secured; e-SISTAFE rolled out to Ministry of Finance, Education and MPD
UTRAFE develops its first Action Plan and Budget for the period 2006-2008 and implements a new organizational structure.
2006
1st National PEFA Assessment (covering the years of 2002, 2003 and 2004) is concluded.
Autoridade Tributária de Moçambique (ATM) – Mozambique Revenue Authority established.
Roll out of e-SISTAFE to a further 28 central level ministries and agencies
Development of the Budget Preparation Module
Establishment of the Common Fund in support of SISTAFE development
2007
Approval of the Financial Management Manual by Government (Ministerial Diploma 169-2007), an important recommendation of the Tribunal Administrativo.
Roll out of e-SISTAFE to 4 remaining central level ministries
Development of the STA Multi-Currency.
Development of the integrated database for staff and state agents using a census (single source for payroll processing)
2008
2nd National PEFA Assessment (covering the years of 2004, 2005 and 2006)
Budget Preparation Manual for OE 2009 introduces the programme based budgeting approach
Start up of the process for the development of a PFM Vision / Strategy for 2009-2020
Roll out of e-SISTAFE to 21 autonomous institutions at central level and 31 districts; Payment of salaries using direct execution modality in 22 central level institutions
Development of payroll, asset management, revenue collection and internal / external control modules and functionalities
2009
Revised legislation approved for Tribunal Administrativo (Law 26/2009)
Roll out of e-SISTAFE to provincial level institutions
Pilot of the use of the programmatic classifier in 3 sectors (Public Works, Education and Agriculture)
Conclusion of the staff census (102 sectors and 32.969 individuals registered)
2010 Procurement Regulations revised (Decree 15/2010)
CEDSIF is established (merge between CPD
UTRAFE converted into CEDSIF and respective regulations approved
Roll out of e-SISTAFE to 31 central and provincial
43 Replacing the Regulamento da Fazenda the financial management regulations in use since 1901 !
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 103
Date Key Event / Development in PFM (including IFMIS development from 2004 onwards)
and UTRAFE) level institutions and 21 districts
e-Folha is further development and concluded
e-Tributação starts to be developed for tax payments
2011
Mozambique adheres to EITI - the dialogue and debate around Natural Resources gains momentum;
A new Public Enterprises Law is approved by Parliament; A law for PPPs, Large Scale Projects and Concessions is passed;
Tax legislation is reformed and updated;
3rd National PEFA Assessment (covering the years of 2007, 2008 and 2009)
PFM Vision document submitted to the Council of Ministers (covering 2011-2025)
Roll out of e-SISTAFE to 78 institutions (but financed directly by such institutions): 38 central level, 22 provincial level and 18 districts
e-Folha is expanded to 116 central level institutions and 33 provincial directorates
Conceptual model for Planning and Budgeting developed;
Business Case for HR Module also developed.
2012
Preparation for IPSAS is started by Public Accounts Directorate of Ministry of Finance
Development of IPSAS Action Plan
PFM Vision document approved by Council of Ministers (June 2012)
CEDSIF revises its regulations and further develops its organizational set up
106 new institutions operating with e-SISTAFE: 27 central level, 64 provincial level and 15 districts
e-Folha expanded to 595 institutions: 6 central level, 136 provincial level, 104 district level and 349 schools
Automatized NUIT system established
2013 Review of the 2009 GBS MoU
Implementation of IPSAS Action Plan
Roll out of e-SISTAFE to more institutions (70)
e-Folha security functionatilities improved.
Source: Compiled by Mariam Umarji, MB Consulting
198. An IMF mission concluded in 200944 that the level of satisfaction with the reforms
implemented was high and that important benefits had been generated. Missions that
followed (2010 to 2013) confirmed government commitment and the satisfaction of the
stakeholders involved. Weaknesses were also pointed out by the IMF missions – notably
fragilities in PFM not related to e-SISTAFE. These were issues related particularly to the
comprehensiveness of the budget and to internal controls. It was also pointed out that the
budget formulation and execution process was hampered as stakeholder knowledge and
understanding of e-SISTAFE was low. Ministries, auditors, donors and many other key
players lacked adequate knowledge about the functioning of the system. Weaknesses were
also identified in the functioning of e-SISTAFE and in the coordination provided by the
Technical Unit for State Financial Administration Reform (UTRAFE).
199. More recent missions from the IMF (December 201345) reported that progress with
structural reforms continued through several important measures, including:
Approval of the Action Plan for the Expansion of the Electronic Payment System
(including the e-folha) and the Civil Service Integrated Database (e-CAF).
44 Teresa Dabán, Maria Betânia Gonçalves Xavier e Paulo Henrique Feijó, (March 2009) Avaliação Das
Reformas Da Gestão Das Finanças Públicas E Plano De Acção Para O Futuro, IMF Fiscal Affairs Department.
45 Letter of Intent, Memorandum of Economic and Financial Policies, and Technical Memorandum of Understanding, December 2013
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 104
Submission of Integrated Investment Plan to the Council of Ministers in July 2013,
with a commitment to further to ensure the link with the debt sustainability
analysis.
Staffing of a payment systems oversight unit in the Bank of Mozambique (BM)..
200. Regarding future reforms, it has been documented by the IMF that the Government will
continue to implement reforms to strengthen PFM namely46:
Wage Bill Management: The action plan to expand the system for direct salary
payments to civil servants and agents in institutions with direct access to e-
SISTAFE is being implemented (structural benchmark in MEFP-June 2013). The
expansion of the payment of salaries by direct bank transfer to all institutions
covered by e-CAF and with direct access to e-SISTAFE will be completed by March
2014 (structural benchmark). The Government will expand e-Folha to all civil
servants in institutions with direct access to e-SISTAFE (about 90% of all civil
servants) by end-2014 (structural benchmark).
Expenditure Execution: The roll out of e-SISTAFE has reached 69 per cent of all
spending units (including at the district level). By end-2014 e-SISTAFE should reach
75 per cent of spending units and a full rollout is expected by end-2015. On July 1,
2013, the commitment component of e-SISTAFE was activated which will help to
improve tracking all phases of the expenditure chain, from commitment to
payment.
International Accounting Standards: a proposal for a new legal framework has
been prepared (Normas e Plano de Contas) to meet international public sector
accounting standards based on IPSAS. This proposal has been submitted to public
consultation and will be submitted to the cabinet for approval in the first quarter
of 2014. The implementation of this framework will require the approval by the
Parliament and the subsequent revision of the SISTAFE Law, both expected by end-
2014. The new accounting standards and the required adjustments to e-SISTAFE
software are expected in 2015 in order to allow for the pilot implementation of the
accrual accounting based on the IPSAS in 2016.
Economic Classification of Revenue: a new revenue classification has been
prepared and is expected to be used in the 2015 budget, in line with best
international practice.
201. In November 2013, the IMF also conducted a mission for the introduction of the Fiscal
Transparency Assessment in Mozambique. This assessment plans to look into 3 key areas
46 The Staff Report for the First Review Under the Policy Support Instrument and Request for Modification
of Assessment Criteria, January 2014
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 105
set out in the new Fiscal Transparency Code: Fiscal Reporting, Fiscal Forecasting and
Budgeting, and Fiscal Risk Analysis and Management. Once the report is published and
recommendations are discussed and agreed they should be properly integrated into the
reform programme.
Impact of reforms on the quality of PFM systems
202. The Public Expenditure and Financial Accountability (PEFA) methodology provides an
internationally accepted approach to the assessment of PFM systems. During the
evaluation period, this methodology has been applied 3 times (2006, 2008 and 2010). The
figure above groups the results of the 31 indicators of the PEFA into 8 categories in order
to summarise the performance of the PFM system as presented in the 3 PEFA
assessments47.
203. The most significant improvements over the period have been in the following areas:
Budget aggregates: the deviation between overall approved budget and the
respective execution is reducing over time and in the case of revenues, execution
now consistently surpasses the forecasted amounts (Indicators PI-1 and PI-3);
Comprehensiveness and transparency of the budget is improving: the budget
documentation presented to Parliament has improved significantly (Indicator PI-6);
Tax administration (access to, compliance with and control of taxpayers) and
revenue management (including the collection of revenues and respective transfer
and reconciliation with Treasury) have improved significantly (Indicators PI-13, PI-
14 and PI-15).
Better use of budget classifiers at budgeting, execution and reporting stages
alongside the generation of more consistent and regularly available information
(Indicators PI-5 & PI-10);
The introduction of and gradual roll out of the electronic salary payment system
(e-Folha) is contributing to better payroll management and control performance
(Indicator PI-18); and
47 Indicators are grouped as follows: Credibility of the budget – PI 1-4; Comprehensiveness &
Transparency – PI 5-10; Policy-based budgeting PI 11&12; Revenue collection & management PI 13-15; Cash management PI 16 & 17; Payroll, procurement & internal controls PI 18-21; Accounting, recording and reporting PI 22-25; External scrutiny and Audit PI 26-28. Indicators of Donor Practices have been excluded. The graphical presentation is derived by assigning a numerical value (from 1 to 7) to each possible score in the PEFA methodology (D, D+, C, C+, B, B+, A) and then averaging across each area of assessment
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 106
Accounts reconciliations are regularly and timely undertaken due to the use of
direct budget execution in e-SISTAFE combined with the introduction of CUT
(Indicator PI-22).
Figure 21: Comparison of PEFA Scores for Mozambique - 2006, 2008 and 2010
204. Performance has deteriorated in two areas:
Credibility of the budget, at the level of the composition of expenditure out-turn by
ministry or institution (Indicator PI-2); and
The quality of Legislative scrutiny of the annual budget law has fallen due to the
fact that the MTEF (CFMP) is not submitted to Parliament meaning that scrutiny of
medium term government policies is weak (Indicator PI-27).
205. Finally, Indicator PI-23 related to “Availability of Information on Resources Received by
Service Delivery Units” has performed poorly in all the three assessments scoring a D,
largely because a PETS survey has not been undertaken. (It was also a PAF indicator for
many years and later dropped from the matrix).
The relevance of the current reform programme
206. There are six strategic objectives set out in the 2011 – 2015 PFM vision document:
Objective 1: Establishment and modernization of the programme based approach
to planning and budgeting and accounting alongside the decentralization of budget
execution to the lowest level of budget holders of the system;
Objective 2: Increase internal revenue and savings and adjust expenditure patterns
to the internal resources available;
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 107
Objective 3: Improvements in the state asset management capacity as well in the
management of government shareholdings and interests in companies; proper
management of public debt and of social security;
Objective 4: Proper accounting and reporting from all budget holders and state
institutions regardless of their autonomous status;
Objective 5: Decision making processes based on proper data generated through
trustworthy management information systems with consequent improvement in
internal control, monitoring and reporting;
Objective 6: Modernization of processes, systems and technological infrastructure
based on the training and valorisation of national human resources.
207. In the strategy, the leadership of reform coordination and management is given to
CEDSIF (Centro de Desenvolvimento de Sistemas de Informação de Finanças – the Centre
for Development of IT Systems, resulting from the merger between UTRAFE and CPD, the
former Data Processing Centre of the Ministry of Finance.) Yet, the details on how this will
be operationalized are missing. Moreover, it is not clear whether it is appropriate (or
feasible) for a unit primarily responsible for the IFMIS (e-SISTAFE) to lead a set of reforms,
which are much wider in their coverage. This links back to the IMF recommendation (IMF,
2009) to clarify who within the Ministry of Finance should coordinate and manage the
overall reform programme.
208. The PEFA assessments and other available documentation clearly demonstrate that
substantive improvements have been achieved in the quality of PFM systems. Nevertheless,
there are major challenges that lie ahead and it is not clear that an adequate framework
exists to address these challenges. In particular, we note the following:
The 2011 – 2015 PFM Vision document presents an ambitious vision. Yet, it is of
necessity a complex reform programme and it is not clear that the reform ideas as
presented are sufficiently developed to be confident of effective implementation:
o The sequencing of reform actions remains incomplete. Although this can
be addressed, to a degree, during implementation it presents real
challenges, for which well-qualified advisory support will be needed.
o A specific problem of sequencing is presented by the need for clear
conceptual designs to precede the development of IT solutions. While this
is not always possible and it can sometimes be advisable to pilot potential
solutions before finalising the conceptual design, there are areas of
reform where IT solutions are being developed before conceptual models
(and the related framework of rules and regulations) have been tested and
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 108
finalised. This is the case, for example, in relation to programme budgeting
and also with regard to the geographical decentralisation of budget
execution, where the issues related to decentralization vs. deconcentration
of execution have yet to be clarified.
o There is also a problem with the scope of reforms, whose focus is on
central government rather than on the broader public sector. In the new
era of extractive industries, there has been a proliferation of public
enterprises, joint ventures and PPPs, new types of autonomous funds and
institutions and other forms of government business, which generate
major risks of actual and contingent liabilities. There is a need for more
comprehensive attention to the monitoring of fiscal risks, within a “whole
of government” perspective.
In relation to the coordination of reforms, two problems remain. The first is over
the appropriateness from a technical perspective of CEDSIF as the reform
coordinator, and the second is over the level of authority CEDSIF can exercise, in
trying to monitor and coordinate the actions of other departments with a higher
level of seniority and status. The coordinator of reforms should ideally have a
wider perspective on PFM issues and a higher status, in order to exert authority.
Regarding the advice and support available through dialogue with the G-19, two
questions arise. Firstly, can adequate technical expertise be marshalled to give
useful guidance on reform sequencing and the other more complex technical
issues? Secondly, can this advice be provided in a coherent, integrated and timely
way?
209. To a degree, each of these problems has been present in the past during the
implementation of the government’s PFM reforms. As Mozambique moves to more complex,
“second generation” reforms, it becomes more important to address these weaknesses in a
systematic way.
The contribution of Budget Support to PFM reform
210. The TA and capacity building support provided to PFM has been essential to the
success of the reform programme and without doubt, the central place of PFM in the
Budget Support dialogue has had a positive impact on the pace and content of reform:
There has existed throughout the evaluation period a clear PFM reform
programme, to which Government has shown a demonstrated commitment.
Nevertheless, the prominent role of PFM reform in the Budget Support dialogue
and in the PAF process have influenced the pace of PFM reform and the direction:
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 109
in particular, stakeholders agree that, in the absence of GBS, external and internal
audit would not have been accorded the degree of attention they received, and
that the roll-out of the e-SISTAFE system would probably have been slower.
The joint analytical work developed by the G-19 PFM working group has also
contributed to improvements in the focus and design of the PFM reform program.
Also important have been the technical assistance and direct financial support to
PFM reform provided through the common funds for e-SISTAFE, for the ATM
(Autoridade Tributária de Moçambique – Mozambique Revenue Authority), for the
IGF (Inspeção Geral das Finanças – the Finance Inspectorate or Internal Audit
service) and the Tribunal Administrativo (TA – the Supreme Audit Institution).
Budget Support dialogue provided the framework in which to identify needs and
track progress, creating a strong synergy with capacity building through these
common funds, which themselves were initiated through the Budget Support
process (See Chapter 3.4).
4.3. Changes in the composition of public spending
211. Evaluation Question 3.3 aims to assess changes in the composition of public
expenditure, in sector policies, and in the resulting composition of spending outputs, and to
examine to what extent these changes may be attributed to Budget Support. In this section,
we examine changes in the economic and sectoral composition of spending. In section 4.4,
we address the second part of the question, presenting our analysis of the evolution of
policies in agriculture, education and health. In chapters 6 and 7, we then link this analysis
to the Step Two analysis of the outcomes and impacts of the agriculture and education
sectors.
Influence of Budget Support on the economic composition of spending
212. In section 3.1, we examined the principal changes in the economic composition of the
budget, as illustrated in the overview of Central Government Fiscal Operations presented in
Table 16. Over the evaluation period (2005 – 2012), we noted that:
Total spending increased by some 9 percentage points of GDP, as a consequence of
both expanded recurrent spending (approximately 5 percentage points) and
expanded development spending (approximately 4 percentage points).
Domestic revenue collection expanded dramatically during the period rising from
14.1% of GDP in 2005 to 23.5% in 2012. Income and profit taxes were the most
important component boosting this performance.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 110
Recurrent expenditures stabilised at 19.4% of GDP in 2011, and have since 2010
been fully covered by domestic revenues.
Within recurrent expenditures, there was an increase in the relative weight of
wages and salaries, which rose from 6.8 % of GDP to 10.2 %, although allocations
for goods & services and transfers also rose modestly.
Annual interest payments increased marginally from 0.8 % to 1.0% of GDP
213. These changes in the composition of the budget have served to protect development
spending, while maintaining recurrent spending at a level that could be sustained from
domestic revenues. A significant part of the increase in salary spending has been due to
the expansion of staff in priority sectors, notably health and education but salary increases
have also been significant. With the salaries budget exceeding 10 % of GDP in 2012, it has
reached a level which is high by international standards. Allocations to transfers and
subsidies (4.3 % of GDP in 2012) have also been relatively high, in part due to the fiscal
demands of the government’s fuel subsidy. Thus, while development spending has been
protected, this has been more a consequence of fast revenue growth rather than control of
recurrent expenditures.
214. Issues related to the economic composition of spending do not receive significant
coverage in the PAF. The G-19 Budget Analysis Group does undertake analysis of this issue,
which is considered during Annual Review discussions. For example, comment has been
made on the rising salaries: GDP ratio. Nevertheless, the evidence does not suggest that
budget support-related dialogue had any additional influence on government policy in this
area over what would in any case have been achieved through the IMF’s Policy Support
Instrument.
215. On the other hand, Budget Support funds were important in facilitating the shift in the
composition of spending towards domestically financed development expenditures, which
grew from 3.2 % of GDP in 2005 to almost 6 % in 2012. The special contribution of Budget
Support comes through its ability to enhance discretionary resources, that is the pool of
resources which are not pre-assigned to any specific expenditure and whose use is at the
full discretion of Government, subject to the relevant laws and regulations and the wider
policy commitments of the Government to the PARPA and to the MoU.
216. How can we be confident that Budget Support resources facilitated the expansion of
domestically financed development spending? A simplistic analysis of the allocation of
Budget Support might suggest that, as it has comprised on average 15 % of total spending,
it has contributed 15 % of each category of spending, and, as such, its contribution to
development spending has been no greater than its contribution to interest payments.
However, such an argument ignores the true nature of budgetary decision-making
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 111
processes. If there had been no Budget Support flows at all, the full requirement for
interest payments would need to have been paid, as well as other constitutionally
protected items of expenditure. It would also have proven difficult to cut salaries by 15 %.
In short, GoM would have been forced to finance the non-discretionary items of spending,
while cutting discretionary items, such as domestically financed development spending.
217. For most of the evaluation period, it is Budget Support, which has funded domestically
financed development spending, as well as contributing to some recurrent spending prior to
2010. Figure 22 provides an estimation of the likely contribution of Budget Support to
budget funding, taking into account the relative discretionality of different categories of
expenditure. It illustrates that for much of the evaluation period, domestic revenue would
only have been sufficient to cover recurrent spending. This trend changed in 2010, by which
time the impressive growth achieved in domestic revenues made it possible to finance all
recurrent spending and a considerable proportion of domestically financed development
spending.
Figure 22: The contribution of Budget Support to the funding of discretionary spending
Source: Fiscal tables from the Bank of Mozambique (for years 2005-2011) and *IMF Article IV (2012) 1/ Budget Support, Basket Funds and Projects include concessional loans and grants; 2/ Domestic borrowing and non-concessional external borrowing are not shown.
218. As the scale of Budget Support has declined, its relative significance as a funding
source has fallen from its peak of nearly 20 % of the total of funding sources in 2007 to
just over 9.4 % in 2012 (Table 20). Domestic revenue has expanded its importance over the
period to comprise more than 71 % of funding in 2012, while the relative importance of
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
2005 2006 2007 2008 2009 2010 2011 2012*
Composition of Central Government expenditure in comparison with sources of funding (% GDP)
Devpt Foreign Financed
Domestically financed devpin non-priority sectors
Domestically financed devpin priority sectors
net lending
Other
Transfers
Goods & Services
Salaries
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 112
Basket Funds and (externally financed) Projects has also decreased but not as sharply as
that of Budget Support.
Table 20: Composition of Budget Funding (excluding non-concessional borrowing)
2005 2006 2007 2008 2009 2010 2011 2012
Budget Support
16.64% 18.28% 19.61% 16.83% 16.33% 14.81% 12.46% 9.42%
Basket Funds
10.36% 9.72% 11.73% 11.16% 10.38% 8.89% 7.56% 6.41%
External Projects
18.41% 16.08% 6.10% 13.72% 6.04% 13.61% 16.52% 12.93%
Domestic Revenue
54.59% 55.92% 62.56% 58.29% 67.24% 62.69% 63.46% 71.24%
Source: Fiscal tables from the Bank of Mozambique (for years 2005-2011) and *IMF Article IV (2012)
Influence of Budget Support on the sectoral composition of spending
219. Total spending on the priority sectors designated in PARPA has more than quadrupled
in nominal terms over the evaluation period, increasing by slightly more than 7 percentage
points of GDP. (Table 21.) As a percentage of total expenditure, education, health,
agriculture, good governance and infrastructure, together with the other smaller priority
sectors, have increased their share from 61% to just over 67% of total spending. Thus,
budgetary allocations – boosted by Budget Support disbursements – have been consistent
with the planned expansion of priority sectors outlined in PARPA II and PARP48.
Table 21: Evolution of Spending within Priority Sectors during evaluation period
Priority sectors as % of nominal GDP
2005 2006 2007 2008 2009 2010 2011 2012
Education 4.85% 4.88% 5.74% 6.23% 6.26% 6.31% 6.58% 6.48%
Health 3.32% 3.36% 3.66% 3.20% 3.14% 2.61% 2.70% 3.81%
Agriculture & Rural Dvt. 1.16% 1.10% 0.91% 1.03% 1.37% 1.18% 2.55% 2.55%
Infrastructure 2.90% 4.05% 3.56% 3.94% 3.81% 4.88% 5.65% 5.20%
Good Governance 2.00% 1.96% 2.01% 2.60% 2.87% 2.67% 2.41% 2.54%
Other priority Sectors 0.21% 0.23% 0.24% 0.39% 0.41% 0.37% 0.46% 1.05%
Total Priority Sectors 14.44% 15.58% 16.13% 17.39% 17.86% 18.02% 20.34% 21.62%
Priority sectors as % of total expenditure
2005 2006 2007 2008 2009 2010 2011 2012
Education 20.56% 18.77% 20.86% 22.47% 19.26% 19.91% 18.27% 20.13%
Health 14.08% 12.90% 13.29% 11.53% 9.67% 8.24% 7.49% 11.84%
Agriculture & Rural Dvt. 4.90% 4.24% 3.30% 3.71% 4.21% 3.73% 7.08% 7.93%
Infrastructure 12.29% 15.57% 12.95% 14.20% 11.71% 15.40% 15.68% 16.16%
Good Governance 8.49% 7.54% 7.32% 9.37% 8.83% 8.44% 6.68% 7.90%
Other priority Sectors 0.88% 0.88% 0.89% 1.41% 1.25% 1.17% 1.28% 3.25%
Total Priority Sectors 61.20% 59.90% 58.60% 62.69% 54.93% 56.89% 56.47% 67.21%
Sources: Ministry of Finance and Fiscal tables from the Bank of Mozambique (for years 2005-2011) and *IMF Article IV estimates (2012)
220. The level of spending on the priority sectors has not been one of the indicators steadily
48 However, the definition of priority sectors in the poverty reduction strategies and in the budget
documents is not fully consistent.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 113
repeated within the PAF. Its use as a PAF indicator was curtailed firstly because of the
difficulties in relating government allocation priorities stated in the PARPA II and PARP to
the budget, and secondly, due to a desire to focus more on the results of spending.
However, allocation to priority sectors has been a subject of discussion within the Annual
Review processes and within some specific working groups (e.g. the Budget Analysis group).
Figure 23: Actual allocations to priority sectors compared with estimated allocations of Domestic Revenue, showing the funding gap filled by Budget Support (MZM Trillions)
Sources: Own computation based on Ministry of Finance data.
221. How may we measure the contribution of Budget Support funds to the growth of
spending in these priority sectors? Again, a simple proportionate contribution to each sector
does not capture the real dynamics of budget decision-making. We have therefore
constructed estimations of the relative contribution of budget support to each sector based
0
5
10
15
20
25
30
2005 2006 2007 2008 2009 2010 2011 2012
Educa on
0
5
10
15
20
25
30
2005 2006 2007 2008 2009 2010 2011 2012
Health(incl.HIV)
0
5
10
15
20
25
30
2005 2006 2007 2008 2009 2010 2011 2012
Agric.&Ruraldevp.(adjustedin2010)
0
5
10
15
20
25
30
2005 2006 2007 2008 2009 2010 2011 2012
Infrastructure
0
5
10
15
20
25
30
2005 2006 2007 2008 2009 2010 2011 2012
GoodGovernance
0
5
10
15
20
25
30
2005 2006 2007 2008 2009 2010 2011 2012
Otherprioritysectors
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 114
on the following assumptions:
All sector ministries and agencies will tend to defend their sector shares of the
budget. Thus, increasing allocations to priority sectors by directly cutting other
sectors is extremely difficult to do. A more likely scenario is that incremental
spending for the priority sectors is sourced from new sources of discretionary
spending, such as budget support.
We have therefore assumed that each sector would continue to command the
same share of domestic revenue throughout the evaluation period. Thus, we have
assigned domestic revenue by sector according to the sector shares ruling in 2004.
Then, the gap between the assumed share of domestic revenue (the black line
based on the above assumption) and the actual share recorded as recurrent and
internally financed development (the area below the green area in each chart in
Figure 23) can be assumed to have been financed by General Budget Support.
222. From this analysis, we draw the following conclusions:
Education is the sector that has commanded the largest share of the budget. To a
significant extent, this has been supported by funding from projects and the FASE
common basket fund (shown by the green area in the chart). However, the 2004
share of domestic revenue would have fallen well short of the funding requirement
for recurrent expenditure and for internally financed development spending. It thus
seems very unlikely that the funding levels provided to this sector would have
been feasible in the absence of Budget Support.
Infrastructure and Health are the next most important sectors in terms of
budgetary allocations. They have also received very substantial external funding.
Indeed, due to the weight of external funding in these sectors, it has been possible
to reduce the share of domestic revenue allocated to these sectors from the share
provided in 2004 and to redistribute these domestic revenue shares to other
sectors. Thus, there has been no demand for Budget Support funding for these
sectors.
The Good Governance sector received a level of allocations in excess of its 2004
share of domestic revenue. Thus, some recurrent expenditures (at the beginning of
the period) and some internally financed development spending (later in the
period) would have been difficult to finance in the absence of Budget Support.
Nevertheless, the overall share of the sector and the estimated influence of
budget support are lower than in the education sector.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 115
A small contribution of Budget Support to internally financed development
spending can be identified in the agriculture and rural development sector, from
2010 onwards. However, the composition of this sector was changed significantly
in 2010 due to the inclusion of rural development and fisheries and this may
distort the applicability of the 2004 revenue share49.
Other priority sectors such as Social Action and Labour have represented less than
4 % of the budget. Nevertheless, the significant increases in spending in 2011 and
2012 exceeded the 2004 share of domestic revenues and would probably have
been more difficult to achieve in the absence of budget support.
223. We conclude that the major contribution of Budget Support funding has been to
support the expansion of public spending in education, and to a lesser extent in good
governance and in agriculture. It seems likely that, in addition to funding internally
financed development spending in the education sector, Budget Support has facilitated the
expansion in salary costs due to the employment of new teachers. Salary allocations in the
sector have increased significantly from 3.1% of GDP in 2005 to 4.7% in 2012, and have
comprised the predominant share of the increment in recurrent funding.
4.4. Changes induced in sectoral policies
169. In this section, we consider the second part of Evaluation Question 3.3, relating to the
changes evidenced in sectoral policies and the relative influence of Budget Support
dialogue upon these changes. Government policies related to macroeconomic management,
to governance and the fight against corruption and to the design of PFM reforms are
considered in other sections; here we focus on policies in three sectors, which feature
within the PAF, namely agriculture, education and health. We first outline the main changes
in policy, considering the respective influence of Budget support in each case, before then
drawing some general conclusions.
Budget Support influence on Policy developments in Agriculture
224. The agricultural sector in Mozambique comprises smallholders, a few “emerging” or
medium-scale commercial farmers, and a handful of large companies that produce, trade,
and process agricultural products, often for export. These companies are mainly involved in
traditional export crops, but recently the mix has diversified with tobacco, bananas, rice,
and sesame. Smallholders account for 99% of the total farms in Mozambique. Producers
organized into small organizations and forums only accounted for 7.2 per cent of farmers
in 2008 (TIA, 2008). Input use (tractors, ploughs, fertilizer, pesticides) is low. At national
49 To avoid this negative effect, the 2004 sector share of revenue has been applied up to 2009 and after
that an adjusted assumption based on 2010 shares of revenue has been applied.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 116
level, only 3.9% of smallholders use fertilizers (CAP, 2010), at an average rate of 4 Kg per
hectare - one of the lowest in Africa. The econometric analysis presented in Chapter 6
examines sector outcomes and their determinants in more detail but, in short, the
agriculture sector remains largely undeveloped and the impact of public policy on spending
on the sector continues to be relatively limited. Why is this and how far has policy adapted
to address these problems?
225. Considering first the evolution of agricultural policy, it is clear that there has been no
shortage of agricultural policies or strategies:
The Agriculture Policy and Implementation Strategy (Política Agrária e Estratégia
de Implementação-PAEI) adopted in 1995 was the first agriculture policy document
since Independence. PAEI aimed to promote food security, sustainable economic
growth, employment and reduction of absolute poverty.
PROAGRI I was launched in 1999 with the objective of improving farmers’ access
to financial and agricultural services and building the capacity of the Ministry of
Agriculture to support sectoral growth and transformation. An important focus of
the policy was on increasing food crop production and promoting diversification.
(Chilonda et al., 2009).
PROAGRI II was adopted in 2005 and was partly aimed at addressing the
weaknesses of the previous programme, which was seen as too centralised and
generally ineffective. PROAGRI II aimed at decentralizing some of the mandates of
the Ministry of Agriculture and emphasized a result-based methodology.
Three new rural development strategies were adopted in 2007 - the Green
Revolution Strategy (Estratégia de Revolução Verde-ERV), the Rural Development
Strategy (Estratégia de Desenvolvimento Rural - EDR) and the Investment Fund for
Local Initiatives (Orçamento de Investimento de Iniciativas Locais - OIIL)50.
In 2008, PAPA, the Action Plan for Food Production 2008-2011 was launched. PAPA
aimed through national programs to concentrate efforts on increasing production
and productivity in certain crops and products considered key for household
consumption.
In 2011, PEDSA, the Strategic Plan for Agriculture Sector Development, was
approved. PEDSA was a more comprehensive agriculture strategy, which
synthesised several existing strategies such as the Green Revolution, the Irrigation
50 The OIIL was established to create jobs, improve food production and promote income generation
activities. (Cunguara, 2011). It is colloquially known as “the seven million” because it was based on grants of Mtz 7 million for each district to be allocated by Members of Parliament. It was initially restricted to rural areas but was subsequently expanded to urban areas too.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 117
Strategy, PAPA, the Research Strategy, the Reforestation Strategy and the
Nutrition and Food Security Action Plan and Strategy. PEDSA also addressed
climate change as another threat to agricultural development. For the first time
also, more attention was given to infrastructure development and linking
production areas to major markets.
Mozambique is one of the countries in the G8 New Alliance for Food Security and
Nutrition, a commitment by G8 members, African countries, and private sector
partners to reduce poverty through inclusive agricultural growth. Under the New
Alliance, the Government of Mozambique and G8 members have endorsed a
country-specific Cooperation Framework, through which the Government of
Mozambique has committed to specific policy actions that will improve the
environment for private investment in agriculture.
Linked to this is the Comprehensive Africa Agriculture Development Programme
(CAADP), an Africa-wide initiative of NEPAD, to which Mozambique has signed up,
developing the Programa Nacional de Investimento do Sector Agrário (PNISA –
National Agriculture Investment Plan) as part of this commitment.
226. Despite the plethora of policy documents and strategic initiatives, the performance of
the sector has fallen below expectations. As we note in Chapter 5, national accounts data
from INE show that there has been growth in the agriculture sector at an average annual
rate of 7.5% over 2001-2005 and of 7.8% over 2006-2011. This is only marginally below
the target growth rate of 8 – 10%, established in PARPA/ PARP. However, the problem lies
more in the source of this growth. Mogues & Benin (2012) point out that the primary
source of growth has been expansion in the area under cultivation and that there has been
no improvement in total factor productivity. Without an increase in land and labour
productivity, sustained growth in agricultural production cannot be achieved. Data show
that levels of productivity in agriculture started from a very low base, rose very slowly
over 2002-08 and then declined thereafter (TIA surveys; Cunguara 2011). As we note in
Chapter 6, this is primarily explained by the very low use of fertiliser, pesticides and
improved seeds.
227. Evaluations of PROAGRI II conclude that it did not have a positive impact on farmers’
income precisely because access to agriculture services diminished, the use of improved
technologies reduced and agricultural productivity declined. (Cunguara, 2011). Cabral
(2009) argues that PROAGRI had little or no effect on sector outcomes because public
sector interventions did not address the challenges facing the agriculture sector. There was
limited attention to the development of input and output markets, to the strengthening of
extension services, and to tackling the low level of agricultural input use and the limited
access to technology.. The PARPA II assessment studies (Cunguara & Kelly, 2009) also
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 118
showed that many of the agriculture goals were not achieved, notably, those related to
irrigation, access to information on agricultural prices, access to extension services and use
of improved seeds and fertilizers.
Table 22: Government Budget Allocations to the Agricultural Sector, 2005-2012
228. What explains this failure of public policy? Some have argued that the overall level of
spending on agriculture has been too low: NEPAD, in particular, calls for a minimum of 10%
of public spending to be dedicated to agriculture. However, the government budget
allocated to agriculture has increased significantly over the evaluation period as shown in
Table 22 above, and comprised nearly 8% of total expenditure by 2012 (Figure 23, Section
4.3).
229. Evidence suggests that the problem lies at least as much in the composition of
agricultural spending. A World Bank public expenditure review51 reported that in 2007,
almost 80 per cent of total spending was undertaken by the MINAG central and provincial
administration, with the Agricultural Research Institute of Mozambique (Instituto de
Investigação Agrária de Moçambique; IIAM) comprising only 4.2 per cent of total recorded
spending. In analysing PROAGRI, Cabral (2009) notes that much of the growth in
agricultural spending was skewed towards investments in institutional development52 and
away from the financing of agricultural services for farmers and support to improve access
to inputs. Mogues & Benin (2012), while asserting that further increases in agricultural
spending are needed, also call for new investments to be restructured towards research &
development, support to farmer services and increased input use and away from general
expenditures on salaries and institutional support.
230. The evidence available suggests that Budget Support processes were unable to have
any significant influence on these identified weaknesses in agricultural policies and
strategies. The evaluation team has not been able to analyse the relevance of the changes
in policy and spending patterns that may have been introduced during 2013. However, data
and documents covering the period up to and including 2012 show that there were a
51 World Bank. Mozambique - Analysis of Public Expenditure in Agriculture, February 2011
52 Institutional development was here defined as comprising increases in staff numbers and salaries, investments in MIS systems, and increased financing of planning and budgeting processes (such as the annual activity planning exercise), which together comprised a high proportion of PROAGRI spending.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 119
number of inherent policy weaknesses at the sector level, which the link to the GBS process
does not appear to have mitigated in any significant way53, notably:
Fragmented national structures for agriculture sector policy making – with many
different strategies and policy initiatives, a large number of potentially competing
institutions and a large number of not always well coordinated Development
Partners;
Deficiencies in monitoring and evaluation systems;
A disconnect between the policy targets in the PROAGRI assessment matrix and the
activity plans and budgets of the Ministry of Agriculture; and
A disconnect between the policy priorities emerging from research and evaluation
and the agriculture sector policy targets in the GBS PAF, which focussed on
irrigation, land registration and public extension. None of the 20 indicators used
over 2005-2012 related to the degree of access or use of fertilisers, improved
seeds or pesticides.
Budget Support influence on Policy developments in Education
231. The Education sector has been guided by a common Strategic Plan since 1999. The
plan focused on increasing access, improving quality and strengthening institutional
capacity. It was the first comprehensive sector plan that sought to guide external support
to the sector under one policy framework. However its focus was predominantly on primary
education. It was followed by the Plano Estratégico de Educação e Cultura (PEEC) 2006-
2011, which was more comprehensive, covering the entire education sector. It continued
the same focus of expanding access and improving education quality but brought in a
strong equity and gender dimension and much needed attention to HIV/AIDS. This plan and
the current Plano Estratégico de Educação (PEE) (2012-2016) were closely aligned with the
PARPA and PARP. The latest education plan prioritizes efficiency and quality, and has made
improving retention and reducing dropouts a major focus.
232. A significant feature of policy development in the sector has been the consistent focus
on harmonisation and alignment. Alignment has focused on ensuring that planning and
budgeting instruments are complementary and consistent, and form a unified and coherent
framework linking global (PARPA/ PARP) and sector objectives, central and local processes,
and short and medium term perspectives (Ernst & Young, 2007). The second, related,
priority has been to channel external funding increasingly through common mechanisms,
and in practice through FASE. Interviewees reported that these two objectives had seen
53 We analyse the processes of policy dialogue more fully in Chapter 3, section 3.2.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 120
important progress – as is also evident from the increasing levels of funding through FASE,
and the decreasing numbers of “stand-alone” projects.
233. One of the achievements in the sector in the evaluation period has been the
strengthening of the dialogue with the Ministry of Finance. Education sector stakeholders
are convinced that in the absence of this strong dialogue, it would have been more difficult
to obtain the substantial additional budgetary resources provided for the recruitment of
teachers. The MoE plans to build on this good relationship in the future to argue for other
priorities areas – such as the Apoio Direito as Escolas (ADE)54 – to be included in the
Government budget in the future.
234. As regards policy change, documentation and key interviews concur that the most
important reform in the education sector took place in 2004 i.e. before the evaluation
period. This involved the introduction of free primary education and a related package of
reforms, including:
Abolishment of tuition and other fees in primary education;
Provision of free textbooks for all primary education pupils;
Increased funding to schools through the introduction of the ADE grants - a
capitation grant that goes directly to primary schools for non-salary expenses;
The introduction of a new curriculum for primary education in 2004.
235. Situated within the period covered by this evaluation are two areas of policy reform,
namely:
Reform of the pre-service teacher training policy for primary level, with 10th grade
entry by candidates, and one year of professional training, replacing the previous
10 + 2 year training model. This reform reduced the amount of time spent in
training and resulted in a more than doubling of the number of teachers that were
being trained per year (from between 3,000 to 4,000 teachers per year prior to the
reform to almost 10,000 per year from 2008, and 8,500 in 2012). The design of
this model was informed by a detailed study, financed by the Fast Track Initiative
(FTI), which analysed the financial implications of different options of teacher
education and sought to identify what would be needed to obtain 10,000
additional primary school teachers per year (Bartholomew et al., 2009). The new
teacher education policy was to be accompanied by expanded in-service training,
although progress in this area has been slower.
54 The Direct Support to Schools (DSS) initiative, which was started in 2003 and has focused on children
in public primary schools (in grades 1 through 7). ADE provides funding directly to all primary schools for essential school supplies, with the purpose of contributing to improved teaching and learning, and
ultimately to improved pass rates, transition and retention rates.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 121
The reform of the school construction strategy to address the challenges that had
arisen with the introduction of a community involvement model (influenced by the
FASE partners) which was based on a per classroom price of $10,000. This model
proved to be difficult to implement in certain areas of the country due to capacity
constraints.
236. These two reforms were part of a concerted effort to expand access to education and
to improve quality of teaching by reducing the pupil teacher ratio. As we note in Chapter 7,
econometric analysis suggests that this focus on reducing pupil teacher ratios has been
absolutely appropriate, showing a strong association with progression rates through
primary schooling and transition rates into secondary school. In the case of school
construction, the increasing availability of flexible funding through FASE permitted
expansion, and this reform may be said to be primarily driven by FASE processes, although
Government funding has increased for internally funded investments, and this includes
school construction (Figure 23, Section 4.3).
237. Implementation of the teacher expansion strategy was facilitated to a significant
extent by the existence of the GBS dialogue. Increasing the number of teachers required
approval of additional recurrent funding for salaries by the Ministry of Finance. Policy
dialogue and reflection at the level of the sector led to the identification of the pupil-
teacher ratio as one of the three key indicators for the overall PAF framework. This was a
deliberate choice by the Ministry of Education, to present a convincing argument for the
need to increase the salary bill. Thus, a combination of sector dialogue, the existence of a
monitoring framework at GBS level, and the availability of funding through GBS provided a
framework that allowed for the Ministry of Education to engage in direct dialogue with the
Ministry of Finance to obtain substantial increases in recurrent GoM funds to expand
teacher numbers.
238. Within the evaluation period, there have also been important efforts to increase age
six enrolment in grade 1, through mobilization and advocacy efforts among parents and
caretakers55. This policy has had positive effects, with the percentage of age 6 enrolment
in grade 1 rising from 36% in 2002 to 72% in 2012. Again this was a policy initiative born
out of sectoral level analysis and debate but where the GBS dialogue contributed to raising
awareness and commitment to this policy objective, both through the adoption of an
explicit PAF indicator on right age enrolment and through discussions at the Annual
Reviews. (See Section 3.2)
55 The age of enrolment is critical in determining future chances for school completion, especially
because children who reach adolescence while still in lower primary have a much smaller probability of completing primary education.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 122
239. A related development in the sector has been in the area of Early Childhood
Development (ECD). The education strategy (PEE) recognizes the importance of ECD and an
ECD strategy paper has been finalized through a consultative process. ECD was also
included in the funding proposal to the Global Partnership for Education (GPE – the former
FTI) submitted in 2010. Roll out of the ECD strategy started in 2014 with an IDA credit
from the World Bank, channelled through FASE. This will involve management of early
childhood facilities by private/ civil society actors. Discussions on expanding GoM funding
for this initiative are on-going and there is the potential for also bringing this subject into
the GBS dialogue on education.
240. In summary, during the evaluation period education policy has been refined, building on
the strong policy and strategy framework already existing. Policies implemented within the
period have significantly improved sector outcomes. In the absence of Budget Support
funding, effective implementation of these policies would have been impossible, without
either significant reductions in the scale of system expansion, or the adoption of funding
models (such as the payment of teacher salaries through projects and common basket
funds), which would have jeopardised the longer term sustainability of benefits.
241. The contribution of policy dialogue through Budget Support has been more modest but
also important. A robust framework of policy dialogue has been built up building on the
SWAP structures, which preceded GBS at the national level, and education sector policy
development has continued to be driven by the FASE process. However, the national level
dialogue associated with budget support has been used to lend additional weight in key
policy areas, notably the need to reduce pupil-teacher ratios through teacher expansion, to
introduce a learning achievement baseline so as to monitor the quality of education more
effectively, to manage the process of budget decentralisation more effectively, and to give
attention not just to Gross Enrolment Rates but specifically to right age enrolment in year
1.
Budget Support influence on Policy developments in Health
The Framework for sectoral policy dialogue: alignment & harmonisation
242. In terms of the framework for dialogue, for some time the health sector has been
guided by a common Strategic Plan (PESS), which is aligned with the PARPA/PARP. In 2007,
in a deepening of the harmonised approach, the partners in the health SWAp, PROSAUDE,
developed a common PAF which currently comprises 35 indicators. It is actively monitored,
and its monitoring forms the key focus of the annual joint review (ACA). A major
development in policy alignment has been the alignment of the JAR and the annual review
of the health sector with the Annual PES, which is the central document for annual
planning. This alignment has been in place since 2008.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 123
243. In addition to the increasingly aligned nature of this modality, PROSAUDE has
encouraged harmonisation through its focus on the acceptance of common audit practices
and common financial reporting systems among participating health partners. Although
there have been several diagnostics, they have been shared widely and have helped to
form a common view on health sector financial management. A third aspect of
harmonisation during the period under review is the reducing frequency of missions from
partners. This has been a Health PAF indicator since 2008, and between 2008 and 2012
the number of individual health missions has fallen from 37 to 9, and the number of joint
health partner missions has fallen from 9 to 5.
244. In the later part of the period under review, the views of health partners have begun to
diverge with differences in several areas including: on the use of modalities (GBS/ SBS,
Common funds, vertical programmes); the emphasis that should be placed on governance
and audit requirements; disbursement conditions and disbursement responses to
governance shortfalls, audit reports, etc.; attitudes to province-specific programmes; and
views on the levels of the government contribution to health. These fragmented views have
proved to be a major distraction. They have reportedly made it more difficult than
previously to develop common DP positions and have lowered the productivity of dialogue.
Key health sector policy objectives and changes within the evaluation period
245. In broad terms, health policies have remained remarkably consistent. Policies to
strengthen Human Resources in Health, to combat key diseases and HIV/AIDS, and to
address infant and maternal mortality, have continued in modified forms under revised
strategy documents. A central pillar of health policy since independence has been
universality. User fees are accepted as a means to broaden the resource base of the sector,
but they have an adverse impact on access, which is a key goal. Universality is also
compromised by “under the table” user fees. The goal of universality continues, although a
recent report (Cumbi et al, 2012) notes that there is a changing ideology accompanying the
growth of private health provision, whereby health is seen less and less as a free public
service and entitlement.
246. Although much has remained the same, there have also been notable developments.
Decentralization and deconcentration policies have changed the way that the health sector
operates. Provincial Departments of Health have their own budgets, and are empowered by
e-SISTAFE to commit expenditures from allocations in the Single Treasury Account (CUT-
Conta Única do Tesouro). However, a recent report (Cumbi et al, 2012) has argued that in
this process the SDSMAS units at primary level have suffered because they receive funding
from the District Secretariat and compete with other priorities.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 124
247. The period under review is largely covered by two poverty Reduction Strategy
documents PARPA II, which covered the period 2006-2010, and the more recent PARP which
covers the period from 2011. The PARPA II has four key components: for health:
improvement of access to health care services, priority programs, improvement of quality,
and increase in efficiency and cross-cutting topics (gender and inter-sectoral cooperation).
It prioritises initiatives in several key areas of health including: reduced maternal mortality;
prevention of vertical transmission of HIV/AIDS; numerous other HIV/AIDS interventions
including a major expansion of TARV; reduction of infant and child mortality through
greater attention to childhood diseases and expanded immunisation; improved nutrition;
action on endemic diseases especially malaria; reduction of the prevalence of leprosy; and
reduced morbidity and mortality rates from TB.
248. The PARP (2011-2014) is a much more concise document than its predecessor. Like the
PARPA II it seeks to promote equity in access to health care, with special attention to
health and nutrition for women, children and other vulnerable groups. It also retains a
focus on reducing the impact of major epidemics such as malaria, tuberculosis and
HIV/AIDS, intestinal parasites, diarrhoea and other pandemics; it has a greater emphasis on
nutrition, and seeks explicitly to reduce chronic malnutrition and protein-calorie and
micronutrient malnutrition; a third focal area is the improvement of human resource
management, humanizing services with emphasis on quality and on satisfying users’ needs;
the fourth and final focal area is the prevention and mitigation of HIV/AIDS.
249. There is general agreement within the sector that the health sector PESS is fully
aligned with PARPA II and the PARP of 2011. The programmes and goals indicated in
poverty strategies are all included in the 35-indicator PAF of the sector and are therefore
monitored through the JAR. It is difficult to identify the extent of influence of the budget
support dialogue on the alignment of health sector policy with national policy. However, the
PARPA II and PARP are products of national dialogue processes associated with budget
support, and the shared language and shared objectives suggest that the two areas of
dialogue – sectoral and national - have been mutually reinforcing.
Policy implementation: changes in the composition of spending to better reflect health policy objectives in PARPA and PESS
Figure 24: Estimates of Per Capita Health spending by Province, 2010 in Meticais
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 125
Source: Bibi Umarji, 2011
250. The nature of the budget classification system and the significant level of off-budget
spending make it difficult to examine the policy orientation of spending. However, we
consider two aspects of spending composition to determine whether expenditure reflects
national objectives. The first of these is equity in distribution of funds amongst provinces.
Equity in distribution of resources is a major goal of the PESS and PARPA II. Both place
great stress on the equitable expansion of access and the latter asks for a review of the
criteria for allocating resources.
251. Available estimates suggest a highly unequal distribution of health resources amongst
provinces. For example, estimates for 2011 shown above indicate that the per capita
allocation for Sofala was 3.4 times the level for Manica. The allocation of resources by
province is tracked in the Health PAF (Indicator 23) and the structure has barely changed
since it first started to be estimated in 2005. To the extent that access to health is
dependent on more equitable distribution of resources for health care, there would appear
to have been limited progress towards this PARPA and PESS goal.
252. However, there are major shortcomings in the data as presented. In the first place, the
two largest referral hospitals in the country are located in Maputo and Sofala (Beira) and,
at least in principle, serve populations much larger than their own inhabitants. Indeed, the
majority of hospitals, training schools and research institutions serve regional or national
populations and yet tend to be concentrated in the larger urban centres. Secondly, and
perhaps more importantly, available expenditure data simply record spending in relation to
the location of budget responsibility, not in relation to the location of service delivery:
many purchases, such as medicines, are managed centrally or regionally but distributed
much more widely. Thirdly, even if it were possible to identify expenditure on primary
health care and basic curative services by location, there are many reasons – related to
population densities, transport costs, differences in the incidence of diseases – why it
109 118 143 155 160 161
238 263
320 326 371
0
50
100
150
200
250
300
350
400
Per Capita HealthSpending inMeticais
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 126
would be inappropriate to assume that an equal allocation would represent an optimal
solution. In short, there remains much work to be done to develop appropriate indicators
for this policy objective before meaningful targets (and, thus, corresponding actions) can
be properly defined56. As the budget classification develops and evolves, it may prove more
helpful in the short to medium term to develop indicators and targets based on data from
the Demographic & Health surveys and/ or Household Budget Surveys.
253. The second aspect of spending composition is the split between the government
budget, Common Funds (PROSAUDE I & II) and vertical funds. Again, the weaknesses in
data availability, especially for off-budget expenditures, complicate the compilation of
accurate estimates57. Nevertheless, Table 23 suggests that from 2005 to 2011 the total
health budget almost doubled in US$ terms. The government contribution to health rose
from $104 million to $160 million, but fell as a proportion of health spending from 30.6%
to 24.2%. Common Funds fell from 31.2% to 15.2% and vertical funds rose from 38.2% of
spending to 60.6%.
Table 23: Aggregate Health spending by source of funds, 2005 and 2011
Source of Funds 2005 2011
US$ million %
US$ million %
Government Budget 104 30.6 160 24.2
Common Funds 106 31.2 100 15.2
Vertical Funds
130 38.2 400 60.6
Total Expenditure 340 100.0 660 100.0
Source: Evaluation team estimates, based on data from MISAU.
254. The majority of the vertical funds budget is for HIV/AIDS through PEPFAR, leading to an
imbalance in programme funding. Estimates indicate that spending on HIV/AIDS in 2011
and 2012 exceeds 40% of the health budget. HIV/AIDS is clearly and necessarily a priority
area, but it is not the only one. There is nothing in the PESS or other documents that would
support such a heavy bias of spending in its favour. However, PEPFAR has plans to move
resources into health systems strengthening and this might go some way to redress the
balance. This trend away from more aligned modalities and towards vertical funds is
clearly not influenced by Budget Support, although without PROSAUDE II the movement to
non-aligned modalities would be more marked.
56 Some government stakeholders in MISAU and MPD raised these concerns with the evaluation team. It
was not fully clear to what extent these concerns are appreciated by the health sector DPs.
57 A particular problem relates to the management of end-of-year project balances: with the Government Budget (OE), all balances remain in the CUT and new spending needs to be authorized in the new Budget and, as a result, estimates of spending are quite accurate. By contrast, off budget vertical projects accumulate balances from one year to another and will thus tend to overstate spending if estimates are based on new transfers into the account, without netting off existing balances.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 127
Policy implementation: achievement of health sector outcome targets
255. During the period under review 60% of health targets in the GBS PAF were achieved
and a further 21% were not achieved but showed progress. Only 17% were not achieved
without progress. (See Figure 15 in Section 3.2.) Reduction in HIV transmission has been a
major focus and numbers of children under ARV treatment stood at 25,891 in 2012, a
dramatic increase from 2005 (1,686). For adults the corresponding figures were 282,687 in
2012 (2009:156,688). Prevention of mother to child transmission of HIV saw a major
increase so that in 2012 85,587 women were treated (84%) compared with only 8,244
(6%) in 2005. The number of health facilities providing PMTCT services rose to 1,170 in
2012 (2005:96). The ratio of institutional births – a consistently reported indicator within
the GBS PAF as well as the Health PAF – has also improved significantly from 49 % in
2005 to 64 % in 2012.
256. In short, across the majority of the outcome targets, which feature in the Health PAF
and/ or the GBS PAF, significant progress has been recorded, indicating some degree of
success with health policy. Three caveats need to be expressed, however. Firstly, as we
noted in Chapter 2, the legacy of the civil war continues to influence health care outcomes
in Mozambique, which tend to be lower than other SSA countries of similar per capita
incomes, such as Tanzania, therefore future progress will need to be still faster if this gap
is to be closed. Secondly, there is a big question mark over how efficiently resources are
being used because, at least at first sight, the progress in outcomes does not appear
commensurate with the big budget increases obtained. Thirdly, there is legitimate concern
over the sustainability of the health benefits being achieved, given that the level of
external funding for the sector has risen so fast.
Conclusions on health policy development and the influence of Budget Support
257. In summary, some positive changes in spending composition and in health outputs
have resulted from positive synergies between GBS and other modalities. Budget support
thinking has guided the development of PROSAUDE II and may yet guide it to the
development of SBS. Moreover, as we noted in Section 3.2, in the health sector there is no
doubt that GBS dialogue and sector dialogue are mutually supportive and that national
level dialogue associated with budget support has provided additional weight in key policy
areas.
258. However, Budget Support is only one influence among many and there is no doubt that
the major changes in spending composition over time have reflected non-GBS factors,
especially the rise of the vertical funds. This has had two direct impacts: first, the vertical
funds, especially PEPFAR, have created a massive focus on a specific disease - in this case
HIV/AIDs - changing the spending composition in its favour; and secondly, through
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 128
increased emphasis on HIV/AIDS, PEPFAR has redirected the wider health sector effort in
that direction. Naturally, positive results have been achieved, but as we have noted there
are concerns about the efficiency and, more significantly, over the sustainability of this
spending, and the related results.
259. The big increase in external funding to the sector, as a result of the combined weight
of the vertical funds and the PROSAUDE common basket fund has had a negative influence
on Government funding to the sector. The response to this has been to reduce GoM
contributions to health (as a percentage of budgetary resources) in order to maintain an
appropriate balance between this and other national priorities. In the process, the GBS
resources available to the Government for use through the Budget have been dedicated to
other sectors, as we noted in Section 4.3 above. This has led to sustained dependence of
the health sector on aid, and a further aggravation of the imbalance in funding since the
proportion of disease-dedicated funds has risen and the proportion of flexible funding has
fallen.
Conclusions: overall influence of Budget Support on Sector Policies
260. In assessing the influence of Budget Support on sector policies, only two of the three
Budget Support inputs, namely funds and policy dialogue, are relevant. Technical assistance
and capacity building at the sector level were not provided through Budget support but
rather through sector specific projects or SWAp arrangements, such as PROAGRI, FASE and
PROSAUDE. By contrast, in the areas of PFM (Section 4.2) and governance (section 4.5), all
three Budget Support inputs have been present.
261. As we noted in Section 4.3, GBS funding has been of considerable importance in
Education and, to a lesser extent in Agriculture. In Education, the scale of GBS funding
available had a clear impact on the scope of education policy – in particular, the scale of
system expansion, which has been undertaken, could not have been contemplated,in its
absence. In Agriculture, the direction of policy has been less consistent and coherent and
we cannot identify any specific effects of Budget support funds on the scale of policy
ambitions or the chosen direction of policy.
262. With regard to GBS policy dialogue, the case studies of agriculture, education and
health show that GBS processes can reinforce and add value to sectoral policy initiatives
but their influence will always be distinctly secondary to sector level processes. These
sector level processes have their own history and their own dynamics, on which the
influence of the central level GBS PAF is modest at best. Where the sector dynamics have
been favourable58, such as in the case of education, the GBS dialogue has helped to
58 Key aspects of this favourable dynamic in the education sector included a) a clear and coherent policy
direction, supported by the key stakeholders; b) a degree of planning and managerial capacity within
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 129
sharpen the strategic focus and to raise awareness of sector level issues; where the sector
dynamics have been less favourable, such as in agriculture, and health, GBS processes
have had less influence.
263. It may be that a much more substantive engagement in sector policy at the central
level – with a higher profile in the GBS PAF, more time dedicated to analysis and debate,
and supporting TA and capacity-building inputs would generate a greater influence, just as
it has in relation to PFM or governance issues. Yet, central to the success of such a policy
would be i) an adequate level of alignment to Government policy, and ii) an adequate level
of harmonisation across the DPs working at the centre and in the sector. If these conditions
were not already generated by the pre-existing sector dynamics, it would probably be
difficult to create them without a very heavy investment in harmonisation and alignment
processes.
4.5. Governance and the Fight against Corruption
264. Evaluation Question 3.4 aims to analyse whether there have been improvements in the
quality of governance and accountability and to assess to what extent they may be
attributed to Budget Support. We focus on changes with regard to four key aspects of
democratic accountability: (i) the quality of Parliamentary scrutiny of national budgets,
accounts and audit reports; (ii) the quality of media and civil society analysis and scrutiny
of budgetary issues; (iii) the scope, coverage and quality of the audits of the Tribunal
Administrativo (TA – the Administrative Court/ Supreme Audit Institution); and (iv) the
efficiency and the effectiveness of anti-corruption initiatives in combating corruption. We
assess to what extent these changes may be attributed to Budget Support, either
independently or in synergy with other modalities.
Parliamentary scrutiny of the State Budget and of Audit Reports
265. In line with the Constitution, the Assembleia da República – the National Assembly or
Parliament, has the responsibility for scrutinising and voting upon the proposal of the
Executive for the State Budget (OE), and also for reviewing the Tribunal Administrativo’s
report on the General State Accounts (CGE – Conta Geral do Estado). The extent to which
these tasks are undertaken effectively are assessed in the PEFA methodology and
therefore the 3 PEFA assessments undertaken during the period (2006, 2008 and 2011)
give some indication of trends in this regard.
266. With regard to Legislative scrutiny of the annual budget law, the PEFA assessments
the sector, supported by consistency in staffing (i.e. avoiding frequent changes of ministers, managers and technical staff); c) a high level of alignment and harmonization across sector partners; and d) strong analytical and facilitation capacity on the part of the lead donor, namely the World Bank. The Agriculture and Health sectors performed less well in relation to each of these factors.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 130
have scored the process quite favourably, reporting scores of “B+”, “B+” and “C+” in these
assessments. (See Table 24 below).
Table 24: Legislative scrutiny of the Budget - PEFA scores, 2006 - 2011
Indicator Dimension 2006 PEFA Assessment
2008 PEFA Assessment
2011 PEFA Assessment
PI – 27: Legislative scrutiny of the annual budget law
(i) Scope of Legislature's scrutiny A
B+
A
B+
A
C+
(ii) Extent to which the legislature's procedures are well-established and respected.
A A A
(iii) Adequacy of time for the legislature to provide a response to budget proposals.
A A A
(iv) Rules for in-year amendments to the budget without ex-ante approval by the Legislature.
B B C
267. Legislative scrutiny of the annual budget law covers the annual plan (PES), fiscal
policies, the medium term fiscal framework and medium term priorities, as well as details
of expenditure and revenue. The procedures adopted for the review are firmly established
and respected. The Ministry of Finance submits the annual budget proposals to the
National Assembly by 30th, September each year. The draft budget is then distributed to
the six Parliamentary Committees, covering discrete sectoral areas, and to the Comisão
Parlamentar do Plano e Orçamento (CPO – Parliamentary Commission for the Plan &
Budget), which plays an overall coordination role across the committees. The Committees
have two months to review proposals, develop comments and negotiate relevant
amendments with the Minister of Finance and his/her staff. In early December, the budget
is then presented to the National Assembly for their consideration and approval. It is
accompanied by a set of formal comments on the budget proposals by the CPO. The
process of scrutiny of budget proposals by the Legislature is thus well developed and
consistent with international good practice norms, with regard to scope, procedures and the
time allowed for scrutiny, and has thus merited “A” scores against each of these
dimensions in the three PEFA assessments.
268. The one dimension where PEFA scores are lower, and where a decline was recorded in
the latest assessment, is in relation to the rules for in-year budget amendments by the
Executive. The rules regarding such amendments are clearly presented within the SISTAFE
law and its related regulations. These do not allow the Executive to increase the overall
level of expenditure, to reallocate funds between budget entities or to alter the scope and
nature of tax instruments without prior approval of the Legislature. These rules are
generally respected. However, the rules allow for virements within a budget entity and also,
more significantly, allow for the contingency fund to be re-allocated without prior approval
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 131
of the Legislature. Moreover, there is no requirement to report back to the Assembly during
the year on the scope and nature of these reallocations. Thus, the rules do allow for
extensive administrative reallocations59. In its 2006 report to Parliament, the Tribunal
Administrativo identified these practices as a deficiency within the SISTAFE law and
regulations, but they have not to-date been amended.
269. The Tribunal Administrativo is the supreme audit institution of Mozambique. It
undertakes the external audits of the individual accounts of government entities and of the
public accounts (CGE) as a whole. In addition, it has a judicial status, acting as the Tribunal
de Contas (Court of Accounts) in reviewing and reaching final decisions over the audits of
individual departments, as well as imposing fines and/ or recommending criminal
investigations, where relevant. Thus, in its capacity as a Tribunal de Contas, the TA
performs most of the functions performed by a Parliamentary Public Accounts Committee
in a Westminster-based system.
Table 25: Legislative Scrutiny of Audit Reports - PEFA scores, 2006 - 2011
Indicator Dimension 2006 PEFA Assessment
2008 PEFA Assessment
2011 PEFA Assessment
PI – 28: Legislative scrutiny of external audit reports
(i) Timeliness of examination of audit reports by the Legislature (for reports received within the last three years)
A
C+
B
C+
B
C+ (ii) Extent of hearings on key findings undertaken by the Legislature
C C C
(iii) Issuance of recommended actions by the Legislature and implementation by the Executive.
B B B
270. Only the audit of the CGE and the related audit opinion (parecer) of the TA is submitted
to the National Assembly, which provides a highly aggregated picture of budget execution.
In-depth hearings on key findings do take place but these are mainly with Ministry of
Finance officials. A more substantial involvement of the National Assembly in reviewing
departmental audits would require a change in the legal framework for external audit in
Mozambique. It is this dimension which pulls down the overall PEFA score against this
indicator, earning a “C” score for the coverage of hearings by the Legislature. (Table 25.)
271. Despite the limited scope of the audit reports presented for Legislative scrutiny, the
quality and timeliness of its work has generally scored well, meriting consistent “C+” PEFA
scores across the evaluation period. Legislative scrutiny of the external audit report
normally takes place in the first parliamentary session of the year, beginning in late March
59 This is justified in part by the fact that with the Legislature meeting only twice during the year (for 45
day sessions), some degree of flexibility in budget execution is required for the Executive.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 132
or early April. With the audit report on the CGE being tabled at end November, legislative
scrutiny is thus concluded 4-5 months after its receipt by the Legislature. The Legislature
recommends actions to the Executive, as a result of its review of the CGE and the TA’s
report and opinion. There is evidence that some of these are implemented and there is a
procedure for assessing follow-up, when the CGE of the subsequent year is analysed.
272. The main constraints to the quality of legislative scrutiny relate to the limited time
available, the very modest technical support provided to the CPO and the limited technical
expertise of its members. However, technical support has been provided to the CPO by
UNICEF, by the Westminster Foundation and also by GIZ. Stakeholders report that this is
making a positive difference to the scope and quality of the CPO’s work.
Media and CSO scrutiny of budgetary documents
273. Mozambique’s scores in the Open Budget Survey suggest an improvement in the public
availability of budgetary documents. Its score on the Open Budget Index (OBI) rose from 28
% in 2010 to 47 % in 2012. Despite this positive trend, Mozambique remains far from a
desirable level of budget transparency particularly in terms of the quality of information,
although the Ministry of Finance is undertaking reforms, such as the introduction of
programme budgets and the modernisation of accounting standards, which should over
time help to address this.
274. In 2010, the CPO agreed to participate in a Budget Monitoring Forum – Fórum de
Monitoria do Orçamento (FMO) – engaging civil society organizations interested in
monitoring the budgeting process and undertaking related advocacy work. The forum was
established through informal links with the CPO, and its objective was essentially to
influence the government in resource allocation and transparent execution of the budget.
The CSOs participating in FMO include the Centre for Public Integrity (CIP), which has been
perhaps the most regular and systematic of the CSOs performing analysis related to public
expenditures, including budget monitoring at the district and municipal levels.
275. The significant role played by CSOs and independent research institutes has been
noted by a number of researchers. Shikhani (2012) states that ‘Mozambique’s socio-
cultural system is characterised by the existence of many pressure groups. They constitute
a growing critical mass and play a fundamental role in helping to deepen democracy in the
country.’ Freedom House (2010) reports that ‘academic freedoms are generally upheld’ and
notes that ‘local journalists and non-governmental organisations such as the Centre for
Public Integrity have played a crucial monitoring role by investigating and exposing high-
profile corruption cases.’
276. Nevertheless,, there is a generalized perception that, in recent years, government
aversion to critical voices has risen. Friedrich-Ebert-Stiftung (FES) reported that freedom of
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 133
expression in Mozambique is becoming more restrictive particularly outside of urban
centres (2011)60
. The report on the State of the Union prepared by CESC in 2013 indicates
problems of political intolerance and a strong aversion to criticism on matters of
governance by the government61. Several survey-based perception indices support this
conclusion, as we noted in Chapter 2. Finally, several important personalities from the
political, academic and religious arena have publicly voiced their concerns about the state
of governance in Mozambique62.
277. There are thus many informed commentators, who suggest that the level of
harassment and intimidation of civil society has increased in recent years. Has this reduced
the level of research and analysis of budgetary and governance issues? Has it limited the
number of critical voices? CSOs interviewed by the evaluation team held differing opinions
on this, but the majority expressed the view that the space for research and criticism
remained, even though the environment was becoming more challenging. By comparison
with the situation in 2005, when budget information was less easily available and the
volume of research and advocacy work was more modest, we judge that the quality of CSO
scrutiny of budgetary documents has improved over the evaluation period.
Scope and quality of audits of the Administrative Court (TA)
Table 26: Scope, nature & follow-up of External Audit - PEFA scores, 2006 - 2011
Indicator Dimension 2006 PEFA Assessment
2008 PEFA Assessment
2011 PEFA Assessment
PI – 26:
Scope, nature and follow-up of external audit
(i) Scope/nature of audit performed (incl. adherence to auditing standards)
D D+ C C+ C C+
(ii) Timeliness of submission of audit reports to legislature
C B B
(iii) Evidence of follow up on audit recommendations
C B C
278. The PEFA assessments of the quality of external audit show steady improvements
from an initially low base. (Table 26.) The Tribunal Administrativo was established in the
1990 Constitution but only became operational from 2000, and in the 2006 PEFA
60
Friedrich-Ebert-Stiftung & MISA (2011), Barómetro Africano da Media. Moçambique, 2011.
61 CESC (2013) gives examples of destruction of flags of opposition parties in the provinces of Manica and Gaza, the closure of three community radios with programmes focused on monitoring governance, a coordinator of the civil society platform of Barue threatened by the district administrator with expulsion from the district.
62 We refer in particular to Prof. Lourenço do Rosário, Rector of Apolitécnica; Dinis Sengulane, Bishop of Lebombo; Francisco Chimoio, Archbishop of the Catholic Church in Maputo; and Prof. Carlos Nuno Castel Branco of the Institute of Social and Economic Studies (IESE).
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 134
assessment, its operations were scored poorly primarily because of the small numbers of
audits it was undertaking, which limited its oversight capacity.
279. From 2005 onwards, there has been a sharp improvement in the number and the
quality of audit reports, as documented in recent assessments of the role and performance
of the Tribunal Administrativo63
. Since 2004 the number of audits has increased, the
coverage of audits has grown significantly and gains in timeliness in producing and
submitting CGE opinions have been obtained. The increase in the total number of audits
has been especially dramatic from 29 audits in 2004 to over 400 per year since 2009 (See
Table 27.) In 2006, the Tribunal Administrativo expanded the coverage of its operations to
districts and in 2007 it commenced auditing public works. In 2011, it established its first
permanent provincial offices.
Table 27: Audit Performance of the Tribunal Administrativo, 2003 -2013
Year Audits
planned New Audits
Accounts analysed
(incl. stock)
Accounts judged
2003 5 7 226 0
2004 0 29 272 4
2005 60 68 342 42
2006 70 179 148 49
2007 90 360 192 75
2008 320 350 160 41
2009 450 491 420 72
2010 600 600 214 14
2011 500 502 82 36
2012 450 450 89 102
2013 - - 330 157
Source: Tribunal Administrativo, 2013
280. With the introduction of Budget Support, Development Partners started to lend special
attention to strengthening the oversight of public expenditure, as a vital mechanism to
improve government performance and transparency in budget execution. Technical support
for the Tribunal Administrativo was provided by Sweden and Portugal from 2005, and the
volume of technical and financial support has expanded sharply following the adoption of
the Memorandum of Understanding for the Common Fund supporting the Corporate Plan of
the TA (Plano Corporativo - PLACOR 2007-2010 and PLACOR II 2011-2014). Under the
umbrella of this instrument there was a significant increase in the number of qualified
staff in the TA, particularly among the auditor staff. In 2003 there were only 7 qualified
auditors and this number rose to 300 in 2011 (Ximungo Consultores, 2011.) The coverage
63 Eurosis & MGA (2010) Diagnóstico Institucional do Tribunal Administrativo de Moçambique. Maputo,
Agosto de 2010; CIP (2011) O Destino dos Nossos Impostos – A Conta Geral do Estado 2009 e o Parecer do Tribunal Administrativo. Maputo, 2011; Ximungo Consultores (2011) PSGR Case Study Mozambique: External Audit.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 135
of staff costs was secured through the Common Fund. PLACOR also provided for capacity
building and exchange of experiences, through secondments and study tours, as well as
advisory support for the development of the new External Audit Act, passed in 2009.
281. Government budget allocations to the Tribunal Administrativo were boosted by Budget
Support funds during the evaluation period. Figure 24 shows the trends in the composition
of expenditure of the institution. The influence of PLACOR can be identified as the green
area at the top of the chart, corresponding to externally financed Development. Following
the same methodology explained in section 4.3 of this report, the influence of BS funds
may be estimated as corresponding to the area between the black line and the externally
financed funds (green area). Thus, from 2005 and more substantially from 2007, Budget
Support funds have helped to finance the expansion of activity of the TA during the
evaluation period, providing funding for all internally financed development spending and
some recurrent spending.
Figure 25: Allocations to Tribunal Administrativo compared with estimated allocation of Domestic Revenue, showing the funding gap filled by Budget Support (MZM Trillions)
Note: 2011 and 2012 includes Tribunal Administrativo Provincial
Source: Own computation based on MoF data.
282. However concerns have recently been raised regarding the misuse of funds by the
Tribunal Administrativo, as an audit report in 2013 found evidence of financial
irregularities, due to weaknesses in financial and reporting systems and a failure to adhere
to procurement procedures (Deloitte & Touche, 2013). This indicates that regardless of the
significant support that has been provided by development partners, weaknesses still
remain in internal management and financial systems.
283. Despite this, it is clear that the quality and volume of TA reports and audits has
improved and it would have been extremely difficult to achieve these results if the PLACOR
Common Fund and Budget Support had not both been in place. The PLACOR common fund
has provided capacity strengthening and additional investment funding. In parallel, the
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
2005 2006 2007 2008 2009 2010 2011 2012
TRillionsM
ZM
TribunalAdministrivo(includingprovincialTA)
Externaldevp.
Internaldevp.
Recurrent
AssumedSpending(2004)
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 136
scope for additional national budget funding which GBS has provided has been
fundamental – indeed, in financial terms its value may have been double that of the
common fund. (See Figure 24.) In addition, stakeholders agree that the attention given to
the Tribunal Administrativo in the Budget Support dialogue – including the incorporation of
a PAF indicator with targets for the growth in audit coverage – has helped to raise its
profile and its ‘negotiating strength’ in lobbying for support from other parts of
government. Thus, Budget Support and the PLACOR common fund have worked in synergy
to support the improvement in the coverage and quality of external audit, although it
appears that insufficient attention has been given to strengthening the financial
management systems of the TA itself.
Progress in the fight against corruption
284. The public perception of corruption in Mozambique is high and has remained high
throughout the evaluation period. The report of the APRM (2010) identifies corruption as a
significant problem both at high levels of government through the manipulation of public
procurement processes, and also via petty corruption, which is practiced through bribes to
employees in sectors such as health, and police. Transparency International’s Corruption
Perception Index (CPI) showed a marginal improvement from 2.8 out of 10 in 2005 to 3.1
in 2012 but this change falls within the confidence interval (2.4 -> 3.1) and is thus not
statistically significant. The CPI placed Mozambique 97th out of 158 in 2005 and 123rd in
2012 from 176 countries. Overall, there has been little discernible change in levels of
corruption, rather a continuing and serious problem.
285. Nevertheless, a range of government initiatives have been introduced to strengthen
the fight against corruption:
The new Public Finance Act (the ‘Lei SISTAFE’, Law nº 9/2002) was approved in
2002, setting the framework for new public finance regulations and control
procedures, and for the subsequent development and establishment of the e-
SISTAFE integrated financial management system.
In 2004, Parliament approved Mozambique’s first Anti-corruption Law (Law nº
6/2004).
In 2005 the Public Procurement Law was passed and then subsequently refined in
2010 through Decree 15/2010.
In 2006, the Council of Ministers approved the Anti-Corruption Strategy (Estratégia
Anti-Corrupção (EAC) 2006-2010.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 137
286. In 2009, an evaluation of the implementation of the Anti-corruption Strategy64
concluded that although some progress had been made in implementation, there were
serious shortcomings. It noted that activities had been poorly prioritised and had scarcely
reached the local level. The report also pointed to problems concerning the legal
framework for EAC, which was considered deficient in terms of its inter-institutional
coordination, and also identified severe limitations in the technical capacity and legal
mandate of the Central Office for Combating Corrupting – Gabinete Central de Combate a
Corrupção (GCCC).
287. In March 2010 the government presented a Governance Action Plan (GAP). This was in
response to a communication in December 2009, from the Troika Plus on behalf of the G-
19, advising the government that a breach in the underlying principles of Budget Support
might have occurred. This was prompted in part by the irregularities in electoral procedures
identified by international observers of the 2009 elections and in part by a perception of
inadequate progress in the implementation of anti-corruption measures. It led to the
temporary suspension of Budget Support disbursements by 4 DPs and to a period of
significant tension in dialogue, which became known as the ‘crispação’. (See Chapter 2 and
Chapter 3, section 3.4.)
288. In response, government presented in March 2010 the Governance Action Plan (GAP).
This comprised16 action points aimed at introducing electoral reforms, promoting political
inclusivity, strengthening procurement processes (including for extractive industries
through the EITI), simplifying legal and bureaucratic procedures relating to the private
sector, simplifying land regulations, submitting a package of anti-corruption legislation to
Parliament, and formalising the Government’s response to the recommendations of the
Africa Peer Review Mechanism (APRM) report. The majority of these actions have been
implemented between 2010 and 2013.
289. As part of the Governance Action Plan, a package of anti-corruption legislation was
submitted to Parliament during 2011. Parliament opted to discuss and approve it in a
fragmented manner and 3 of the 5 laws submitted have been passed to date. In 2012 the
Public Probity Law (Lei de Probidade Pública, Law 16/2012) was approved. This law is
applicable to public providers in areas such as morality, transparency, impartiality, conflicts
of interest and respect in the management of public affairs. As a result, a set of prominent
Frelimo party members were eventually forced to resign from their multiple positions,
following a concerted period of pressure from the media and civil society.
290. Two other laws within the “Anti-Corruption Package” were also passed by Parliament in
2012. One established mechanisms to protect the rights and interests of victims,
64 ACS (2009), Avaliação da Implementação da Estratégia Anti-corrupção (2007-2009). Maputo, UTRESP.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 138
complainants and witnesses (including “whistle-blowers”) in criminal proceedings, and
created the Central Office for the Protection of Victims. Another changed the legal
mandate of the GCCC so that it may press charges in corruption cases, while legislation on
the declaration of assets by public servants was also passed. Two laws addressing the
Penal Code have not yet been passed into law. Of those laws that have been passed,
implementation is still deficient and slow progress has been made on achieving mutually
agreed plans and indicators, although, as part of the GAP, measures have also been taken,
since 2010, to strengthen the budget and the staffing of the Central Office for Combating
Corrupting – Gabinete Central de Combate a Corrupção (GCCC).
291. There is evidence of an increased number of corruption cases going to trial. As shown
in Table 28, there has been an increase since 2008 in the number of cases processed by
the GCCC and the number of people accused, and since 2009 in the number of cases going
to court. Overall, the percentage of processed cases which have gone to court has
increased from 9.5% in 2008 to 16.7% in 2012. However, several commentators expressed
concerns to the evaluation team regarding the degree of politicisation of the judiciary
system, suggesting that the potential for fair, unbiased court cases was still low.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 139
Table 28 : Processing of Corruption Cases & Cases of Misuse of State Funds or Property: 2005-2012
Year Cases processed Number accused Court cases
2005 377 13 0
2006 310 35 5
2007 371 19 Not available
2008 619 151 59
2009 623 178 43
2010 649 223 88
2011 677 214 81
2012 526 190 88
Total 4,751 1,155 425
Source: GCCC
292. As a result of the Governance Action Plan and earlier measures, the legal and
institutional framework for combating corruption has been strengthened during the
evaluation period, although progress on implementation has been slow. This has been a
consequence both of actions to strengthen the basis for Anti-Corruption work and broader
reforms to strengthen procurement and public financial management65. Although there is
no evidence of positive, statistically significant change in the public perception of
corruption, the legal and institutional framework to prevent corruption occurring and to
prosecute identified cases of corruption is stronger.
Influence of Budget Support on changes identified
293. Good Governance is one of the three sectors that appear as the primary beneficiaries
of Budget Support funds during the evaluation period. Figure 26 shows the composition of
spending in the sector as a whole and the estimated influence of Budget Support flows,
following the assumptions adopted in section 4.3. In the sector as a whole, just as in the
case of the Tribunal Administrativo, one of its component institutions, a significant part of
the national budget allocations can be linked to Budget Support funds. All in all, it can be
said that without the flow of funds provided by Budget Support and to a lesser extent by
pooled funds and projects, the changes made in the sector would not have been easily
undertaken.
65 See section 4.2 above on PFM, where we report steady improvements in the quality of the system as
measured by the PEFA assessments. These include improvements in the mechanisms for procurement, which have scored respectively “C”, “B”, and “B” in the 2006, 2008 and 2011 assessments.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 140
Figure 26: Allocations to Good Governance compared with estimated allocation of Domestic Revenue, showing the funding gap filled by Budget Support (MZM Trillions)
Source: Own computation based on MoF data.
294. With regard to the “institutional effects” of Budget Support processes (as opposed to
the “flow of funds effects”), three avenues of influence can be identified:
The joint agreement and monitoring with government of targets in the
Performance Assessment Framework (PAF), reviewed in the Budget Support Annual
Reviews.
The engagement of civil society and the media within this PAF review process, and
the sharing of information to assist in the research and policy advocacy activities
of CSOs.
The monitoring of adherence to ‘underlying principles’ and the related policy
dialogue between Government and the G-19.
Figure 27: Good Governance indicators in the PAF (left) and overall performance (right)
Source: Own computation based on MPD data.
295. Governance indicators have been the most prominent PAF indicators but their numbers
0
2
4
6
8
10
12
14
2005 2006 2007 2008 2009 2010 2011 2012
TRillionsMZM
GoodGovernance
InvestExt
Invest.Interno
Funcionamento
AssumedSpending(2004)
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 141
have declined as alternative dialogue channels have been used. Over 2005 – 2012 ,
governance indicators (Justice, decentralisation and public sector reform) have comprised
24 % of all PAF indicators, as compared with 17 % for PFM. However, their numbers have
declined from 2010 onwards. (Figure 27.) In part, this reflects the increased attention given
to underlying principles and to “areas for special attention” in the GBS dialogue; and in
part, the difficulty in using quantitative indicators to drive progress in a sector as complex
as governance.
296. Governance indicators have performed positively during the period, with 48 %
“achieved” and 33 % “not achieved but with progress”. Among them, Justice indicators are
the ones that on average performed most poorly. The relevance of governance indicators,
as tabulated from the opinions of Focus Group members from Government and DPs, scores
quite highly (2.9 out of 4). Moreover, there was a reasonable level of consistency between
Government and DP stakeholders, as to which indicators were considered more or less
important. For example, both parties gave low rankings to the relative importance of
indicators of “corruption cases detected”, “cases judged by year” and “crimes solved”,
presumably because there was a mutual appreciation of the unreliability of the
corresponding data and of the poor link between these data and performance.
Table 29: Performance and relevance of PAF indicators for Governance, 2005 -2012
Sector Key Indicators
Relevance of indicators
based on the Focus Group
poll
PAF Assessment (1=Achieved; 2=Not Achieved but with progress; 3=Not achieved); 0=Not
assessed)
Govt (Ave)
DPs (Ave)
2005
2006
2007
2008
2009
2010
2011
2012
AC
Survey about good governance, corruption and service delivery.
3.67 3.13 3
Anticorruption legislation package 3.67 3.56 1
Number of corruption cases detected
2.33 2.88 1 1 1 1 1 1
J
Cases judged by year 2.67 2.50 1 1 3 3 2 3 3
Penal Sector Reform 3.33 2.93 3 2
Crimes solved 2.33 2.29 2 3 0 2
TA* Coverage of Tribunal Administrativo audits
3.00 3.44 1 1 1 1 1 1 1 1
D
Decentralization policies and strategies
3.25 2.56 3 2
Number of municipalities (autarquias)
2.75 1.92 1
Intergovernmental transfers 3.25 2.94 1 1 1 1 1 1
PSR Wage and salaries harmonization policy
3.00 3.00 3 2 1 2 3 2
*Note TA indicator is not considered as a Governance but a PFM indicator in the PAF AC = Anti Corruption; J= Justice; D= Decentralization; PSR= Public sector Reform; TA=Tribunal Adminsitrativo Regarding relevance, 1= least important, 4 = most important, as judged by Focus Group members.
Source: Own computation based on data from Annual Reviews and responses from Focus Groups
297. Our impression from stakeholders – which is supported by the ‘relevance ratings’
above – is that the PAF indicators provided a useful vehicle, around which to debate and
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 142
review governance issues. They may also have helped to raise the profile of institutions
such as the TA. However, there is no clear evidence that the inclusion or exclusion of a
governance indicator had a direct effect on the degree of political and administrative
effort dedicated to its achievement. For example, progress on justice issues was no faster
with more PAF indicators and no slower without them: this is simply an area, where
reforms are inherently difficult. Thus, in the governance area in particular, the mere
monitoring of PAF targets proved insufficient to generate progress: in general, this has
required additional actions, undertaken either through technical assistance linked to
Budget Support or through more detailed processes of dialogue, such as the matrices
developed for ‘areas for special attention’, including the Governance Action Plan (GAP) and
the payroll reform plan.
298. The public nature of the Annual Review and the use of press statements to announce
its results are perceived as important in promoting accountability. However, CSO
stakeholders did not believe that their interventions in the Annual Reviews made a material
difference to policies. The involvement of civil society was welcomed by CSOs, as an
opportunity to obtain information and participate in debate but targeted advocacy work
and direct dialogue with Government were seen as the most effective ways of making
progress on specific governance questions.
299. There is evidence that the high level debate between Government and the G-19 at the
time of the “crispação” was influential in focusing attention and speeding up the
implementation of actions included in the Governance Action Plan (GAP). Government
stakeholders asserted that all of the issues included within the GAP were already under
internal discussion and would have been implemented but they agreed with DPs that the
high-level attention to these issues helped to speed up their implementation.
5. Growth & Poverty Reduction
300. This Chapter, together with Chapters 6 and 7, reports on Step Two of the analysis,
comprising Evaluation Questions 4.1 and 4.2. These questions examine the main outcomes
and impacts, generated during the evaluation period. Using different quantitative and
qualitative analytical techniques, the EQs seek to identify the key determinants of change
over the period. In particular, they assess the relative significance, amongst other factors,
of the changes in Government’s allocative and policy decisions, which were identified in
Step One as having been induced by Budget Support processes.
301. We present the analysis of Step Two in three chapters, before then drawing our overall
conclusions on the links with Step One in Chapter 8, which comprises Step Three of the
evaluation. Our Step Two analysis covers three areas:
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 143
Economic growth and poverty reduction, considering trends in income and non-
income poverty during the evaluation period;
Agriculture sector outcomes and their determinants, looking specifically at the
relationship between poverty and agricultural productivity; and
Education sector outcomes and their determinants, where we examine the
relationship between the educational inputs, on which budgetary resources have
been concentrated, and the outcomes achieved in primary and secondary
education.
5.1. Growth data and trends by economic sector and province
302. Mozambique has experienced high and sustained growth during the evaluation period
at an average rate of 7.3%, but its progress in poverty reduction has been less impressive.
Although two different methodologies have been applied in measuring and analysing
poverty (the official poverty figures, presented in the Third National Poverty Assessment
and by the World Bank in Alfani et al 2012), both tell a consistent story at the national
level: there has been no statistically significant change in poverty between the early and
the late 2000s. Although the evaluation period covers the years 2005-2011, a broader
perspective helps contextualize better the growth performance in the 2000s.
303. The national growth performance is summarized in Figure 28 using GDP at constant
2003 prices. This shows that while growth has remained positive and sustained over time,
its pace has slowed in the evaluation period with respect to the previous years. The recent
years though have been characterized by a greater stability in growth rates, probably
signalling the end of the stabilization process following the end of the civil war in 1992.
304. The sharp decrease in growth in 2000 reflects the severe floods that hit the country
causing major agricultural losses, as well as destroying existing irrigation infrastructure
and the cultivated land. There was a quick recovery in 2002, after which time growth has
been more stable, at an average of 7.3% in the evaluation period.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 144
Figure 28: GDP growth (constant prices) Figure 29: Growth in selected sectors
Source: Authors’ calculations using INE data Source: Authors’ calculations using INE data
305. Figure 29 shows the performance of three selected macro-sectors, namely agriculture,
manufacturing and mining. The impact of the 2000 floods on agriculture is clear here,
while other sectors proved more resilient. The sectoral disaggregation also reveals the high
volatility of growth in the extractive sector that experienced some very high growth peaks
(2000 and 2004) and some negative growth rates (1996 and 1999). Mining was the fastest
growing sector in the Mozambican economy in 2011. Finally the growth rate in the
manufacturing sector decreased in the last part of the period but it was also more stable
than in earlier years, in line with the overall GDP trend.
306. Table 30 presents average growth rates for all sectors over 5-year periods, the last
one corresponding to the evaluation period. All sectors show decreasing average growth
rates in the last period, except for construction and agriculture (where the increase is
slight). However, average growth rates hide the large intra-period variations shown in
figures 28 and 29. As noted above, the lower pace of growth in the evaluation period is
associated with higher stability in growth rates, which may be a positive feature.
Table 30: Average growth in macro-sectors at the national level
Sector 1995-2000 2001-2005 2006-2011
Agriculture 5.8 7.5 7.8
Mining 15.8 25.6 14.3
Manufacturing 17.1 15.2 3.5
Utilities and public services 8.0 9.7 7.1
Construction 17.3 6.7 8.4
Services and trade 5.7 8.2 7.7
GDP 8.3 8.8 7.3
Source: Authors’ calculations using data from INE
307. It is worth noting that the manufacturing sector has been underperforming during the
evaluation period, being the slowest growing sector in the economy. In particular in 2011
the sector registered a 3% growth in real terms, compared with an overall GDP (constant
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 145
prices) growth of over 7%. This is a reverse situation with respect to the first half of the
2000s when the sector was performing better than any other sector, excluding mining.
308. An analysis of the composition of GDP by the same sectors as in Table 30 shows the
dominance of agriculture and services and trade, together accounting for at least two
thirds of GDP over the period 1995-2011 (39% and 27% respectively of GDP in 2011). The
mining sector, while growing, remained very small, only 1.45% of GDP in 2011. The
sectoral composition of GDP was remarkably stable over this period, aside from a decrease
in the share of agriculture and corresponding increase in manufacturing between 1995 and
2000; after 2000 the sectoral composition of GDP changes little. Agriculture remains a
very important sector in Mozambique, and this is highly relevant for poverty (70% of the
Mozambican population lives in rural areas66) and for overall growth dynamics.
309. GDP data at provincial level is available from 1997 and therefore the analysis at this
more disaggregated level focuses on a slightly shorter time horizon. Table 31 shows that,
following the same pattern observed at the national level, GDP growth has stabilized in all
provinces during the evaluation period (and sometimes slowed).
310. Growth rates have been very similar in all provinces in the last period, at between
6.3% and 8.3%. In this period the slowest growing provinces were Tete and Maputo
Province and the fastest was Gaza. In particular Tete only caught up with others in 2009,
after having experienced a drop in growth due to the 2008 weather shock that particularly
affected the central provinces.
Table 31: Mozambique - Average growth rates by province
Province 1997-2000 2001-2005 2006-2011
Niassa 4.1 8.2 7.2
Cabo Delgado 5.0 7.9 7.7
Nampula 7.3 7.6 7.6
Zambezia 4.1 8.3 7.4
Tete 18.1 7.9 5.2
Manica 5.4 7.8 7.9
Sofala 6.8 8.5 7.4
Inhambane 4.9 13.3 7.6
Gaza 1.1 9.9 8.3
Maputo Province 8.4 12.0 6.3
Maputo City 9.5 7.1 7.8
Source: Authors’ calculations using data from INE
311. The provinces in the table are ordered roughly from North to South, to facilitate the
identification of possible geographical patterns. However there is no obvious north/ south
divide in terms of economic growth, particularly in the last years considered.
66 Estimate from the 2008/2009 household budget survey (IOF 08/09).
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 146
312. An analysis of the sectoral composition of GDP at the province level shows that in
2011 the most agricultural areas were Zambezia and Cabo Delgado (where agriculture still
contributed over 50% of GDP), followed closely by Niassa and Gaza. Manufacturing is
highly concentrated in Maputo Province, where about half of the total national
manufacturing value added was produced. Maputo City reports a high concentration of
services. Finally mining activities were largely located in Inhambane in 2011, whereas in
other provinces they are very small (though due to the lumpy character of mining activities
however, this geographical picture may change substantially even in the span of a year).
Table 32: Consumption, GDP and agriculture growth between survey years
Growth 96/97 to 02/03 Growth 02/03 to 08/09
Province Consumption GDP Agriculture Consumption GDP Agriculture
Niassa 64.8 20.2 6.2 45.1 9.1 10.5
Cabo Delgado 7.6 23.3 10.0 12.7 40.6 38.8
Nampula 18.4 39.3 14.0 10.4 27.7 27.0
Zambezia 34.3 30.2 16.7 -26.9 33.8 30.7
Tete 56.1 73.7 17.0 14.1 -4.6 -30.3
Manica 10.1 18.4 2.3 -15.7 26.8 24.1
Sofala 154.0 42.1 14.9 -19.8 34.9 33.9
Inhambane -10.2 26.7 3.8 74.0 92.4 58.5
Gaza 34.4 13.6 0.2 -12.2 49.6 44.2
Maputo P. 1.7 44.0 -8.1 11.2 15.5 50.2
Maputo C. 19.7 48.1 15.8 4.5 44.9 44.8
National 27.4 38.2 11.0 2.1 31.6 30.7
Source: Consumption from IAF 96/97, IAF 02/03, IOF 0809; GDP and Agriculture from INE GDP series.
313. Table 32 above shows that, at the national level, the growth in per capita consumption
has slowed dramatically over the second period while GDP growth has remained of similar
magnitude. However agricultural per capita production has grown much faster between
2002/03 and 2008/09 (44.8%) than in the previous period (15.8%). This high agricultural
growth, combined with the predominantly rural population, might have been a driver for
poverty reduction. However the survey data show that this growth has failed to translate
into significantly higher consumption (and so presumably lower poverty) in rural areas in
particular, a picture consistent with the poverty estimates presented above. At the
provincial level the growth (both GDP and agriculture) are not significantly correlated with
changes in poverty, perhaps confirming the missing link between growth and poverty
reduction that has emerged throughout the discussion of poverty in Mozambique. Indeed
the correlation coefficients between growth in household consumption and in GDP and
agriculture are respectively 33% and 8%, neither being statistically significant67. It is
notable that the regions with worse poverty reduction performance between 2002/03 and
67 Correlations coefficients are calculated for growth rates at the provincial level between 2002/03 and
2008/09.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 147
2008/09 (i.e. Zambezia, Manica and Sofala) have registered high growth rates in per capita
GDP and agricultural value added.
314. While we might expect growth in consumption and in GDP to be broadly consistent, it is
important to note that they are not expected to display exactly the same pattern.
Consumption is only a component of GDP, which also includes investment amongst others.
While GDP growth matters for poverty, household consumption dynamics are almost
certainly more relevant.
5.2. Trends in consumption poverty by Province
315. This section is concerned particularly with trends in poverty rather than with the level.
While poverty lines and part of the poverty debate is focused on the 13 spatial domains
used to compute poverty lines, here a disaggregation by province is used, which
corresponds to the administrative decentralized structure of Mozambique.
316. Table 33 reports changes in poverty between survey years, using three standard
measures of poverty: the poverty headcount (or P0), the poverty gap (or P1), and poverty
severity (or P2) measures68. As well as confirming the above national patterns of change,
these three measures show that while progress was made in both Northern and Southern
areas, the Central region of Mozambique was poorer in 2008/09 for all the poverty
indicators than in 2002/03. The indicators by Province show that indeed the central
provinces of Zambezia, Manica and Sofala have experienced the highest increases in their
headcount poverty (P0) between the last two survey years (statistically significant for the
first and third; see TPNA) notwithstanding the good performance of 96/97 – 02/03, and
have experienced an increase in both the depth (P1) and the severity (P2) of poverty over
the last period. Other small increases in poverty were registered in Nampula and Gaza but
these changes are not statistically significant.
Table 33: Percentage Changes in poverty headcount (P0), poverty gap (P1) and squared poverty gap (P2), 1996/97 – 2008/ 09
Changes in P0 measure Changes in P1 measure Changes in P2 measure
96-97 – 02/03
02/03 – 08/09
96-97 – 02/03
02/03 – 08/09
96-97 – 02/03
02/03 – 08/09
National -15.3 0.6 -8.8 0.7 -5.3 0.7
Urban -10.5 -1.9 -7.0 -0.6 -5.0 0.0
68 The poverty headcount simply measures the share of the population that reports a level of
consumption below the poverty line for the corresponding spatial domain. The poverty gap measures the gap between consumption and the poverty line as a proportion of the poverty line for all poor households (i.e. those below the poverty line). It indicates how distant is the average consumption level of the poor from the poverty line, thus indicating the depth of poverty. The poverty severity (or squared poverty gap) measures the average of the square of the relative poverty gap, again calculated only on households below the Poverty Line. By giving more weight to households that are more distant to the poverty line, it captures the severity of poverty and it takes into account inequality changes amongst the poor.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 148
Rural -16 1.6 -9.0 1.3 -5.2 0.9
North -11 -8.8 -7.1 -2.9 -5.0 -0.9
Centre -28.3 14.2 -16.7 8.3 -10.1 5.1
South 0.7 -9.6 2.3 -7.0 2.1 -4.6
Niassa -18.5 -20.2 -14.3 -3.5 -9.4 -0.2
Cabo Del. 5.8 -25.8 1.8 -10.1 0.4 -4.7
Nampula -16.3 2.1 -9.1 0.5 -6.0 0.5
Zambezia -23.5 25.9 -12.0 13.9 -6.2 7.8
Tete -22.5 -17.8 -12.7 -9.8 -7.2 -6.4
Manica -19 11.5 -7.4 4.3 -2.5 1.9
Sofala -51.8 21.9 -38.5 16.3 -27.8 12.8
Inhambane -1.9 -22.8 3.6 -21.3 4.6 -15.9
Gaza -4.5 2.4 -2.4 7.7 -1.6 7.4
Maputo Pr. 3.7 -1.8 3.3 -5.5 2.5 -4.7
Maputo City 5.8 -17.4 4.4 -9.1 2.6 -5.1
Source: Third National Poverty Assessment
317. An analysis of inequality (not presented here) shows that nationally and in many
provinces inequality fell, though there were large increases in some cases. The central
provinces identified as poor and poorly performing, particularly Zambezia and Sofala,
registered not only high but also increasing inequality between the last two survey rounds.
However the most dramatic increases in inequality occurred in the north of the country and
particularly in Niassa and Nampula. The two best performing provinces in terms of the
reduction of inequality between 2002/03 and 2008/09 were Cabo Delgado and Tete that
also registered high rates of poverty reduction over the same period.
318. It is possible to disaggregate the reduction in consumption poverty between two points
in time to say how much reflects change in average consumption (growth component) and
how much is due to change in inequality (redistribution component). Table 34 reports the
relative contribution of these two components, calculated using the Datt and Ravallion
(1992) decomposition method. The second column of the table reports the total variation in
the headcount poverty between 2002/02 and 2008/09, calculated for each region using the
averaged poverty line of the corresponding rural and urban spatial domain. The third and
fourth columns report respectively the contribution of the growth and redistribution
components to the total variation.69 The figures in Table 34 indicate that poverty reduction
was mainly due to growth rather than to redistribution, except in Cabo Delgado, Nampula
and Gaza where the latter component is larger; inequality rose in Nampula (associated with
the increase of poverty) but fell in the other two cases (where poverty fell).
Table 34: Growth-redistribution decomposition of changes in poverty from 2002/03 to 2008/09
Province Total variation Growth Redistribution
69 The sum of the two components is not equal to the total variation because a residual is also
estimated using the Datt and Ravallion (1992) method
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 149
Niassa -0.115 -0.201 0.043
Cabo Delgado -0.209 -0.038 -0.159
Nampula 0.082 -0.014 0.100
Zambezia 0.224 0.197 0.030
Tete -0.183 -0.108 -0.054
Manica 0.137 0.170 -0.042
Sofala 0.178 0.195 -0.032
Inhambane -0.268 -0.205 -0.019
Gaza -0.046 0.034 -0.080
Maputo P -0.058 -0.048 0.002
Maputo C -0.192 -0.148 -0.009
Source: Authors’ calculations using IAF 02/03 and IOF 08/09.
319. Finally, growth incidence curves were computed by province and at the national level.
The growth incidence curve gives the growth rates in per capita consumption by percentile,
ranked by the level of consumption. For simplicity only the national growth incidence curve
is reported in Figure 30, and shows a mostly unequal pattern of growth over the period.
Figure 30: Growth incidence curve at the national level, 2002/03 – 2008/09
320. Table 35 reports the rate of pro-poor growth by province, obtained by integrating this
curve up to the headcount poverty index for each province, so giving the mean growth rate
of the poor. This is compared with the growth rate in the mean and in the median. A rate of
pro-poor growth lower than the growth rate in the mean indicates that the poor have
benefited less from growth than the non-poor. This is the case for 5 provinces, including
four that registered the highest increases in poverty: Nampula, Zambezia, Manica, Sofala,
as well as for Niassa (where poverty fell). In general the provinces with the highest rates of
pro-poor growth are also those that performed better in terms of poverty reduction.
Table 35: Rate of pro-poor growth by Province, 2002/03 – 2008/09
Province Pro-poor growth Growth in mean Growth at median
Niassa 0.57 6.40 2.85
Cabo Delgado 3.48 2.01 4.58
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 150
Nampula -2.09 1.66 -1.06
Zambezia -6.3 -5.09 -6.01
Tete 5.70 2.22 4.90
Manica -2.91 -2.8 -2.21
Sofala -5.04 -3.6 -3.02
Inhambane 11.31 9.67 11
Gaza -1.74 -2.15 0.46
Maputo P 4.77 1.79 4.24
Maputo C 6.39 0.73 5.71
National -0.01 0.35 0.93
Source: Authors’ calculations using IAF2002/03 and IOF 2008/09
5.3. Conclusions on Income Poverty in Mozambique
321. The analysis of growth and poverty in Mozambique allows the drawing of four broad
conclusions. Firstly there seem to be a missing link between growth and income poverty
reduction in Mozambique. While the country experienced high and sustained growth rates in
the evaluation period, the poverty reduction performance was disappointing. Indeed the
analysis of pro-poor growth reported shows that at the national level the poor have
benefited less from growth than the non-poor. When growth rates in GDP and in
consumption are compared, this disconnect between the two appears rather clearly.
322. Secondly the sectoral analysis of GDP shows that many sectors experienced good
growth, including agriculture. This sector is particularly important in the context of a highly
rural population (estimated at 70% in Mozambique) and given its large contribution to GDP.
However agricultural growth, that was faster in the evaluation period than before, has
failed to benefit the poor as well. So good agricultural growth in aggregate has not been
enough for poverty reduction.
323. The third conclusion follows naturally from the first two and it relates to data quality.
While the disconnection between GDP, agricultural and consumption growth may well be a
reality, the extent of discrepancies observed require a reflection on the reliability of
available data. National accounts aggregates (GDP and agriculture) include big investment
projects that boost growth but do not necessarily have a strong impact on poverty.
However the counter intuitive patterns do raise questions on the trustworthiness of data.
Both the TNPA and the WB paper highlight the limitations of the current surveys,
particularly at the provincial level, both as regards measurement errors and the long
delays between rounds. An improved survey design and management together with a
higher frequency of survey rounds would surely make the analysis of poverty trends more
reliable.
324. The general poverty story in Mozambique is one of slow progress in the evaluation
period in terms of income poverty. However, it is important to also look at non-income
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 151
poverty indicators, which tend to have lower volatility with respect to consumption. They
capture long-term welfare whereas consumption poverty is sensitive to short term
variations.
5.4. Trends in non-income poverty
325. Turning to non-income poverty, this section considers a set of indicators largely
consistent with those presented in the TNPA for the national level, but disaggregated here
by province. We only consider trends between 2002/03 and 2008/09 with the aim of
exploring whether improvements occurred in non-monetary poverty despite the limited
progress in the consumption dimension.
Table 36: Percentage of households owning durable goods, 2002/ 03 and 2008/ 09
Car Motorbike Telephone Bed
Province 02/03 08/09 02/03 08/09 02/03 08/09 02/03 08/09
Niassa 0.6 0.3 1.3 6.9 0.9 13.0 42.3 53.9
Cabodel. 0.7 0.4 0.8 2.8 0.9 11.5 7.8 75.6
Nampula 0.2 0.9 1.3 6.1 1.4 10.7 31.6 28.3
Zambezia 0.2 0.4 0.7 3.1 0.6 8.2 25.2 21.2
Tete 0.8 0.6 0.5 2.1 0.9 11.2 5.0 7.0
Manica 1.7 1.1 1.4 2.9 2.9 19.9 37.0 24.8
Sofala 1.7 1.4 1.8 3.3 4.0 30.5 28.5 26.1
Inhamb. 2.1 2.8 0.4 1.8 1.4 37.0 48.1 51.0
Gaza 2.6 3.2 2.7 3.2 6.0 45.8 60.2 57.7
Maputo P 3.5 5.8 0.9 2.2 15.2 68.5 69.9 69.0
Maputo C 12.8 11.5 2.4 0.9 42.4 85.3 91.2 85.8
National 1.6 1.8 1.2 3.6 4.3 23.7 34.2 39.0
Radio TV Bicycle Fridge
Province 02/03 08/09 02/03 08/09 02/03 08/09 02/03 08/09
Niassa 43.0 59.6 2.1 5.1 56.9 65.5 1.4 2.3
Cabodel. 43.0 45.3 2.6 4.3 24.1 42.5 2.6 1.8
Nampula 48.3 38.5 3.0 6.0 26.7 35.1 2.2 3.0
Zambezia 39.4 41.3 1.7 3.7 38.7 48.8 1.1 1.9
Tete 45.1 47.4 1.7 3.7 27.9 41.8 2.2 2.5
Manica 63.5 71.3 4.9 6.7 38.5 55.1 4.2 4.0
Sofala 52.3 53.1 7.0 14.6 35.4 44.1 4.8 10.4
Inhamban
e 32.9 42.6 3.9 11.0 11.7 24.0 2.4 4.4
Gaza 34.1 40.3 4.6 20.3 16.7 19.3 7.2 9.5
Maputo P 53.5 43.1 18.3 43.7 10.2 10.2 15.0 26.9
Maputo C 61.8 40.8 56.0 65.4 7.8 5.7 43.3 48.6
National 45.5 45.8 6.3 12.4 28.1 38.1 5.1 7.5
Source: Authors’ calculations using IAF 2002/03 and IOF 2008/09
326. Table 36 presents a set of indicators related to the ownership of a selection of durable
goods. These are the same considered in the TNPA, but are here all disaggregated also at
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 152
the provincial level. The indicators show some improvement in the probability of owning
durables in all provinces, particularly when substitute good are considered jointly. This is
particularly the case for transport durables, namely cars, motorbikes and bicycles, and
information technologies such as radio and TV. However progress has been rather slow,
consistent with the stagnating consumption poverty estimates. This consistency should be
expected in the case of durables, since this aspect is fully in the household’s control
whereas other non-monetary measures, such as education, are also influenced by
government spending.
327. Table 37 looks at indicators related to housing, including sanitation and electricity.
Walls are defined as durable if they are built of concrete or bricks, and the roof is so
defined if built of concrete, zinc, lusalite (fibrous cement) or tiles. Sanitation indicates
whether the household has access to a toilet or improved latrine, while electricity is an
indicator taking the value of 1 if electricity is the main source of energy for lighting. The
indicators show a generally improving trend across provinces, with only a few exceptions
showing a decrease in the reported indicators. None of the provinces experienced a
decrease in all of the indicators. As far as regional differences are concerned, it is worth
noting that Northern provinces seems to be more deprived than Southern ones particularly
in terms of durables (namely phones) and the quality of housing (durable roofs and walls).
Table 37: Percentage of Households with Housing, sanitation and electricity
Durable walls Durable roof Sanitation Electricity
Province 02/03 08/09 02/03 08/09 02/03 08/09 02/03 08/09
Niassa 4.4 6.6 4.4 8.1 4.9 17.0 4.4 5.3
Cabo Del. 3.6 2.2 6.8 11.2 6.0 4.4 3.0 3.8
Nampula 8.9 5.0 8.1 11.9 5.4 6.2 4.9 8.5
Zambezia 2.0 9.5 4.2 11.6 2.9 5.5 2.6 4.1
Tete 8.1 17.2 11.8 15.3 3.1 5.8 3.5 5.7
Manica 13.2 13.3 23.4 25.5 20.4 8.4 6.0 8.1
Sofala 19.7 21.0 37.3 39.1 14.3 24.1 6.7 16.7
Inhamban
e 12.7 13.4 37.8 45.8 5.7 19.2 2.8 3.9
Gaza 17.7 18.2 63.6 67.9 20.2 14.3 6.4 13.1
Maputo P 53.1 67.2 93.6 94.2 27.3 54.0 17.8 45.8
Maputo C 81.6 86.9 99.5 99.8 68.9 79.0 45.7 73.6
National 14.2 17.9 24.8 29.8 11.2 16.0 6.8 13.1
Source: Authors’ calculations using IAF 2002/03 and IOF 2008/09
328. Table 38 takes into account trends in indicators related to literacy and enrolment, at
both the primary and secondary level. Literacy rates are calculated for all individuals
between 15 and 30 years of age. Enrolment rates indicate the share of children of official
schooling age that attend school at a grade corresponding to their age range. Secondary
school includes technical schools. Age groups are defined as: primary 6 to 13, secondary 14
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 153
to 19. Enrolment data show great improvement at both levels of schooling in all provinces,
particularly for secondary schooling.
329. Literacy rates have increased in all provinces except Niassa, but in some cases (Cabo
Delgado, Manica, Gaza) the improvements are very small. In all of these cases increased
levels of enrolment at both primary and secondary level have not translated into
significant improvements in this basic educational outcome. However, the elapsed time
period is probably too short to measure the on-going effects of improved school
enrolments on literacy.
Table 38: Literacy rates (percentage of 15-30 age group) and Education (percentage of children attending school at right age), 2002/ 03 and 2008/ 09
Primary enrolment Secondary enrolment Literacy
Province 02/03 08/09 02/03 08/09 02/03 08/09
Niassa 53.1 74.2 3.0 16.9 39.2 38.8
Cabo
Delgado 62.9 69.1 5.7 9.4 31.4 34.2
Nampula 54.4 66.6 2.2 10.6 34.4 43.8
Zambezia 63.9 76.7 1.8 13.0 40.7 47.6
Tete 57.3 69.3 4.1 13.3 45.5 54.2
Manica 71.2 76.2 6.6 14.3 63.3 64.7
Sofala 66.2 76.6 7.4 22.3 52.5 59.9
Inhambane 79.6 86.7 8.9 18.7 65.9 80.8
Gaza 79.4 82.7 6.7 17.7 65.9 68.8
Maputo P 88.2 90.1 11.4 35.8 79.3 84.3
Maputo C 88.8 92.5 24.9 42.5 88.0 92.3
National 65.9 75.7 6.1 17.3 49.1 56.2
Source: Authors’ calculations using IAF 2002/03 and IOF 2008/09
330. Finally, Table 39 shows three standard anthropometric measures of child malnutrition,
namely:
Stunting, based on height-for-age ratios;
Wasting, based on weight-for-height values;
Underweight, based on weight-for-age ratios.
Table 39: Child malnutrition (% of children two standard deviations from WHO international reference value)
Stunting Wasting Underweight
Province 96/97 08/09 96/97 08/09 96/97 08/09
Niassa 54.7 50.1 10.7 6.0 27.8 19.2
Cabo
Delgado 50.2 54.8 12.8 7.6 36.0 26.6
Nampula 46.8 55.4 13.8 8.2 32.8 23.4
Zambezia 65.9 48.5 4.9 7.5 28.8 19.8
Tete 51.0 51.4 8.9 9.6 34.6 26.5
Manica 60.4 59.2 3.3 4.7 25.8 21.2
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 154
Sofala 54.0 35.7 8.3 11.1 29.7 19.5
Inhambane 34.0 38.6 9.3 2.9 21.2 10.8
Gaza 47.5 34.3 4.7 4.5 15.9 12.4
Maputo P 33.9 20.4 5.4 5.0 13.5 7.6
Maputo C 25.8 25.0 7.2 4.6 10.9 8.2
National 50.6 46.5 8.5 7.1 27.6 19.7
Source: Salvucci (2012)
331. All the variables are measured as the percentage of children that are below two
standard deviations from the international reference value (WHO 2006). The data is from
the household surveys of 1996/ 97 and 2008/ 09 that include anthropometric information
for children. (The 2002/03 round did not collect such information).
332. The figures show that while progress was made at the national level, child malnutrition
is still a major problem in Mozambique. At the provincial level, most regions made progress
in decreasing the prevalence of stunting, except for Cabo Delgado, Nampula and
Inhambane that saw increases in stunting. However some provinces that fared well in
terms of stunting report increased prevalence of wasting. This is particularly the case for
Zambezia, Tete, Manica and Sofala. The underweight indicator has improved in all
provinces. The pattern is therefore far from clear, but it confirms that child malnutrition
remains a major unresolved issue.
333. In general the picture shown by the non-monetary poverty indicators is more positive
than that shown by the monetary poverty measures in particular in relation to education
indicators, and to a much lesser extent for indicators relating to housing and durable good
ownership.
District level welfare assessment
334. So far the discussion has focused on the national and provincial levels. However
provinces in Mozambique cover relatively large areas and intra-province variations in
poverty may therefore be substantial. The household surveys alone do not allow computing
indicators at the next most disaggregated level, i.e. districts, because the samples are only
representative at the national and provincial level. However it is possible to take a closer
look by using census data that was collected in 1997 and in 2007. Two approaches in
particular have been adopted in the case of Mozambique : poverty mapping and first order
dominance (FOD). The poverty maps obtained through the two approaches are reported in
Volume Two.
335. The first approach (Hussain and Castigo 2012) creates poverty maps by district by
combining survey and census data. This uses indicators available in both the census and
the surveys to impute a distribution of poverty at the district level, from which a headcount
measure can be computed and presented on a map. The surveys should be close to census
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 155
years (1997 and 2007) to ensure consistency and indeed this is possible in the case of
Mozambique by using IAF 1996/ 97 and IOF 2008/ 09. The authors use calorie-adjusted
consumption70 to overcome the problem of underestimated consumption for some
households.
336. The results show that consumption poverty has decreased in 115 out of 146 districts
but that differences between districts have increased between 1997 and 2007, thus
pointing to important intra-province dynamics. Districts across the whole country appear to
suffer from high poverty, including in Southern areas where the district level poverty has
become increasingly concentrated. The authors also provide a map to indicate the areas
where most of the poor people live, as opposed to the poverty headcount (which is relative
to the population). Central and Northern areas, and in particular in Zambezia, are home to
most poor people in Mozambique.
337. The second approach is based on the methodology developed by Arndt et al. (2012)
that allows establishing rankings of small areas based on strict first order dominance
(FOD) of welfare indicators in one area with respect to another. Applied to Mozambique
(Arndt et al. 2013), a multi-dimensional measure is used, which includes the following
indicators of deprivation, all available in the 1997 and 2007 census data:
Access to safe water
Access to sanitation (i.e. flush toilet or latrine)
Education (i.e. at least one Household member has some education)
Access to electricity for lighting
Radio (i.e. the Household has a functioning radio).
338. The FOD methodology is used to rank districts by level of deprivation in both 1997 and
2007, as well as to obtain a temporal index indicating advancement in welfare over time.
The results show that the worst-ranked districts are in the Central and Northern areas,
both in 1997 and 2007, while the Southern districts are the best-ranked ones. The
correlation between the FOD index in 1997 and in 2007 is high (0.87), indicating that the
welfare rankings are fairly stable.
339. At the provincial level some changes can be identified, with Zambezia and Nampula
relatively worsening on average and Niassa, Manica and Sofala improving. This confirms
again Zambezia as one of the most problematic provinces in Mozambique. The general
patterns revealed by this analysis are consistent with the poverty mapping approach.
70 Consumption is re-estimated and imputed for households reported to consume less than 800 calories
per capita per day. For details see Hussain and Castigo (2012).
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 156
5.5. Conclusions on Growth & Poverty in Mozambique
340. There seems to be a missing link between growth and poverty reduction in
Mozambique. While the country experienced high and sustained growth rates in the
evaluation period, the poverty reduction performance was disappointing. Indeed the
analysis of pro-poor growth reported shows that at the national level the poor have
benefited less from growth than the non-poor. When growth rates in GDP and in
consumption are compared, this disconnect between the two appears rather clearly.
341. The sectoral analysis of GDP shows that many sectors experienced good growth,
including agriculture. This sector is particularly important in the context of a highly rural
population (estimated at 70%) and given its large contribution to GDP. However
agricultural growth, although faster in the evaluation period than before, has also failed to
benefit the poor. Good aggregate agricultural growth has not been enough for poverty
reduction.
342. The general poverty story in Mozambique is one of slow progress in the evaluation
period particularly with regard to income poverty. Poverty has been assessed here
including both its consumption and non-monetary dimensions. Progress was made in most
provinces in terms of housing and education, particularly in those areas where public
spending is likely to have a large impact such as enrolment rates and electrification.
However slow progress is observed in consumption poverty as well as in malnutrition and
some dimensions of non-monetary poverty, particularly durables (with the exception of
telephones).
343. . With regard to malnutrition and other aspects of non-income poverty the analysis
also points to a disappointing performance in this area. While the non-monetary measures
show a more positive picture on poverty reduction, this change is largely driven by the
improvements in education provision. Chapter 7 presents a more detailed analysis of
educational outcomes, including an analysis of progression, repetition and drop out rates.
6. Agricultural Sector Outcomes and their Determinants
344. The aggregate national accounts data for Mozambique shows that the agriculture
sector has shown significant growth since 1995. Although for much of that period it has
grown at rates slightly below the average for the economy as a whole, it in fact grew
faster than average over the 2006 -11 period. Despite this, when it comes to explaining the
lack of progress in poverty reduction from 2002/03 to 2008/09, much of the blame for this
is attributed to the lack of progress in agriculture, including the lack of progress in raising
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 157
productivity. This is a widely shared concern in Mozambique.
345. This chapter therefore focuses on understanding the trends in, and correlates of,
agricultural productivity in Mozambique over the period 2002 -2008 using data from the
TIA agricultural surveys. The data in particular refers to small and medium farms as these
are the ones most directly related to poverty. Data on community surveys is also used but
only for 2002 and 2003, the only years we were able to obtain. Data from the 2012 TIA
survey was also available, but regrettably did not collect data on production and so could
not be used to estimate productivity.
346. The surveys cover around 5000 farms in 141 districts in Mozambique. The analysis of
productivity conducted here is done at the level of individual farms, as well as at the
community and district level. We begin by presenting some descriptive statistics from the
TIA surveys covering not just productivity but also use of inputs, sale of outputs etc. The
subsequent then presents an econometric analysis of the correlates of productivity, as a
guide to the potential formulation of public policies aimed at improving the productivity of
small and medium farms.
6.1. Trends in agricultural performance
347. The core measure of productivity used here is the value of crop output per unit of
cultivated area, expressed in 2006 province specific prices in order to make the different
provinces and rounds of survey data comparable. A frequency distribution of productivity
across farms in the first and last year for which we have data, respectively 2002 and 2008
shows that throughout most of the distribution, productivity was higher in 2008 than it
was in 2002. This of course can be sensitive to the specific years chosen and climatic
conditions at the time.
Figure 31: Trends in median agricultural productivity, 2002-08 (a) National Pattern
(b) Province level patterns
0,0
100,0
200,0
300,0
400,0
500,0
600,0
2002 2005 2006 2007 2008
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 158
Source: TIA surveys
348. Figure 31 shows the year to year trends in the median productivity of small and
medium farms at the national level and by province. The first panel (a) shows the trend in
average national productivity over the years of the TIA surveys. The pattern is one of a
modest increase in productivity over the period, though with some volatility and actually
declining from 2006 to 2008. The second panel (b) shows the equivalent information at a
province level. This shows both significant variations in the level of productivity by
province, and often different trends over time. Productivity levels are highest in Niassa and
Tete, followed by Cabo Delgado and Maputo, whereas productivity levels are very low in
Zambézia and Inhambane. This is strongly associated with the geographic poverty pattern
revealed by the 2008/ 09 IOF survey: the first three are provinces with relatively low levels
of poverty, the last two are provinces with high poverty levels. In addition, it will be seen
shortly that this is strongly linked to patterns of fertilizer use, which is strongly associated
with productivity and whose use is highest in Tete and Niassa.
349. Niassa and Tete show large increases in productivity over this period, and there are
more modest increases in Manica, Sofala and Cabo Delgado. Other provinces show
0
50
01
00
01
50
0
0
50
01
00
01
50
0
0
50
01
00
01
50
0
2002 2004 2006 2008
2002 2004 2006 2008 2002 2004 2006 2008 2002 2004 2006 2008
niassa cabo delgado nampula zambezia
tete manica sofala inhambane
gaza maputo maputo city
Me
dia
n P
rodu
ctivity (
Mtk
/Ha)
Graphs by Province
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 159
declines (Nampula, Inhambane) or little progress from a low base (Zambézia). An
immediate point to note is the close association between the evolution of productivity and
changes in poverty: Niassa, Tete and Cabo Delgado all had quite large reductions in poverty
over almost the same periods (2002/03 to 2008/09), while Zambézia had an increase. This
association does not however apply to Inhambane where the poverty reduction does not
reflect improved agricultural productivity.
350. The median area cultivated, while not particularly small, does decline over the period
(see Table 40). More than half of the land is allocated to basic crops, and other large
shares to beans and potatoes. The median area allocated to cash crops is small. A minority
of farmers (around 32%) grow cash crops in any case and there is no obvious trend in this.
Table 40: Median Area cultivated in Hectares
Total Basic Food Beans Potato
Cash Crop
2002 1.86 1.16 0.20 0.27 0.10
2005 2.27 1.19 0.28 0.30 0.18
2006 1.93 1.03 0.26 0.23 0.11
2007 1.82 0.96 0.25 0.24 0.09
2008 1.67 0.87 0.23 0.23 0.10
Total 1.87 1.01 0.24 0.25 0.11
Source: TIA surveys
351. In terms of inputs the proportion of farmers using fertilizer and pesticides is very low,
at 3-4% and 4-6% respectively (see Table 41), and it shows no obvious trend over time.
The proportions using manure are also very small. Less than 10% use improved seeds in
most years. In addition only a minority report having access to information; where they do
it is commonly via the radio and very rarely via extension services. But households still sell
quite a lot: the proportion of output which is sold is around 40% which is quite high and if
anything shows an increase over the period from 2003 to 2012. (See Table 42.)
Table 41: Percentage of farmers using different inputs
Improved seeds
Used fertilizer
Used manure
Used pesticides Had
Information Through…
Radio
Extension Service
2002 0.037 0.060 0.068 0.345
2003 0.025 0.018 0.052 0.472 0.315 0.066
2005 0.09 0.038 0.028 0.054 0.403 0.300 0.057
2006 0.14 0.046 0.030 0.053 0.363 0.248 0.033
2007 0.14 0.041 0.048 0.042 0.351 0.207 0.030
2008 0.12 0.041 0.044 0.038 0.341 0.218 0.024
2012 0.10 0.026 0.024 0.060 0.490 0.295 0.023
Total 0.12 0.036 0.036 0.052 0.396 0.263 0.038 Source: TIA surveys
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 160
Table 42: Proportion of output sold, by crop type and year
Sell / Production Share Cash Crop Likelihood
All
Basic Food
Beans Potato Cash Crop
2003 0.40 0.37 0.45 0.59 0.13
2005 0.36 0.35 0.47 0.20 0.63 0.15
2006 0.41 0.34 0.48 0.69 0.14
2007 0.44 0.37 0.50 0.64 0.14
2008 0.44 0.38 0.51 0.66 0.12
2012 0.45 0.37 0.51 0.63 0.14
Total 0.41 0.36 0.49 0.20 0.64 0.14
Source: TIA surveys
352. What is particularly striking is the association between use of fertilizers and
productivity. In aggregate, those using fertilizers have an average productivity more than
six and a half times of those who do not. (Table 43). This pattern is particularly striking in
Niassa and Cabo Delgado as well as in Zambezia, where overall productivity levels are
much lower. The association between use of improved seeds and productivity is much
weaker, with the average productivity of those using improved seeds being 28% higher
than those not using them. Also improved seeds are generally not used more in the
provinces with higher productivity, although Tete is an exception to this.
353. The poor record in getting fertilizers and pesticides to farmers would appear to be
strongly associated with low levels of productivity in Mozambique. An analysis of use of
inputs by decile of productivity shows far higher use of fertilizers and productivity in higher
percentiles, but not much higher use of improved seeds or manure (see Table 44). This
confirms the strong association between fertilizer and pesticide use and productivity.
Table 43: Use of improved seeds and fertilizer and impact on productivity
Productivity by use of improved Seeds
% using improved
seeds Productivity by use of fertilizer
% using fertilizer
Yes No
Yes No
Niassa 1006.94 833.64 0.07 8748.32 455.68 0.113 Cabo Delgado 540.39 641.53 0.04 7454.24 554.66 0.023
Nampula 403.03 348.12 0.06 853.40 352.36 0.028
Zambezia 374.25 247.55 0.08 2270.36 249.36 0.008
Tete 723.20 660.14 0.22 1813.36 416.95 0.172
Manica 392.88 364.52 0.23 853.70 330.84 0.023
Sofala 495.25 352.90 0.11 327.49 335.06 0.009
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 161
Inhambane 105.81 158.79 0.08 145.61 187.65 0.022
Gaza 408.48 352.92 0.15 732.22 328.94 0.034
Maputo 762.97 535.36 0.18 899.70 578.73 0.078
Total 482.26 376.35 0.12 2227.32 335.14 0.036 Source: TIA 2002-2008
Table 44: Productivity and use of inputs by productivity percentile
Percentile Productivity Fertilizers Manure Pesticide Seeds
10th 34.28 0.019 0.051 0.047 0.092
25th 74.49 0.020 0.047 0.040 0.099
50th 311.83 0.040 0.040 0.063 0.113
75th 617.84 0.057 0.046 0.083 0.102
90th 1905.26 0.132 0.054 0.182 0.149
Total 1670.36 0.080 0.047 0.106 0.107
Source: TIA 2002-2008. Seeds only from 2005
6.2. Regression analysis: productivity and input use
354. The main focus of the regression analysis is on seeking to explain correlates of
productivity at the farm, district and community level. Based on pooled data sets of the
2002 to 2008 rounds of the TIA survey, productivity is modeled as a function of farm
specific factors as well as province and year fixed effects. Models are estimated for both
total crop production as well as for specific categories of crops. The key correlates
considered are use of different inputs in production, notably improved seeds, fertilizers,
pesticides, credit and information. The descriptive analysis above already strongly suggests
the importance of fertilizers and pesticides.
355. Table 45 presents results of regressions for productivity at the farm level. This again
confirms that fertilizer use is significantly associated with productivity in aggregate and in
almost all sub-models. Pesticide use and access to credit are each significantly associated
with productivity in the aggregate model. This result does not necessarily imply causality,
as it is equally plausible that more productive farmers choose to use fertilizer or get
access to credit since these factors are determinants of productivity. Use of improved
seeds is positively associated with productivity of basic foods, but the impact is less
consistently significant than for fertilizer. Access to information has a positive association
with productivity of basic foods and potato. A similar model incorporating community level
variables shows that productivity is higher in communities where credit facilities are
available; and similar regressions for productivity at the community or district levels
reinforce the strong association between productivity and use of fertilizer and pesticides.
Table 45: Regression Analysis for determinants of productivity at the farm level
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 162
(1) (2) (3) (4) (5) VARIABLES Total Basic Food Beans Potato Cash Crops
seeds 46.4 400.1** -3,103.9 1,235.6 475.7 (262.696) (159.847) (4,260.529) (930.734) (4,759.278) fert 3,816.1*** 618.9*** 17,464.7*** -1,038.4 13,524.5*** (322.346) (226.075) (5,031.838) (1,491.031) (4,657.434) pest 1,590.3*** -80.6 862.9 -1,744.1 -2,501.4 (280.218) (213.007) (5,226.251) (1,244.080) (3,698.756) estrume 162.1 43.2 8,837.4 -878.5 4,647.4 (386.945) (240.939) (5,811.539) (1,449.955) (7,407.673) credito 1,042.1*** -397.5* -6,668.2 522.2 2,765.8 (351.956) (238.211) (5,837.961) (1,650.368) (4,812.616) precrad -140.1 79.0 -504.6 1,660.5** -930.1 (194.057) (114.244) (3,044.183) (653.250) (3,605.739) precass -313.2 485.8* -2,607.3 1,810.9 -2,931.4 (437.479) (261.765) (6,683.957) (1,631.498) (7,504.848) precext 273.3 673.8*** -3,309.9 -1,325.4 1,542.1 (375.109) (233.652) (6,290.254) (1,130.834) (6,033.375) precpub 485.4 1,013.3*** -1,575.5 -1,349.5 3,307.1 (445.459) (265.951) (7,547.845) (1,578.829) (7,559.079) precout 40.7 -71.6 -2,545.4 -1,683.3* -3,869.6 (218.504) (130.557) (3,455.050) (872.648) (3,981.894) Constant 2,746.6*** 561.9*** 10,584.8** 885.3 45,579.8*** (320.623) (198.406) (4,429.560) (1,338.538) (6,855.028) Observations 8,005 4,717 2,345 610 2,737 R-squared 0.065 0.055 0.033 0.128 0.052 Province FE Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
356. The use by farmers of different inputs as a function of community characteristics (see
Table 46) shows that when credit is available farmers are more likely to use fertilizers and
pesticides, and that where extension services are available and the community is closer to
input sources, farmers are more likely to use improved seeds. However, proximity to the
source of fertilizer is not an important correlate of fertilizer use. Similar models conducted
at the community level emphasize the important association between access to credit and
use of fertilizers as well as pesticides.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 163
Table 46: Regression Analysis of determinants of Productivity and Input use at the Community level
(1) (2) (3) (4) (5) VARIABLES Productivity Seeds Fertilizer Pesticides Credit
distance_fert_time 3.6 0.001 -0.001 -0.001 0.000 (35.446) (0.002) (0.001) (0.001) (0.001) distance_tools_time 10.1 -0.004** -0.000 0.001 -0.001 (39.920) (0.002) (0.001) (0.001) (0.001) credit_any 1,021.5 -0.028 0.049** 0.075*** 0.060** (854.025) (0.031) (0.025) (0.024) (0.024) exten -539.6 0.099*** -0.001 0.028 -0.001 (618.449) (0.022) (0.018) (0.017) (0.017) inf_prices 694.5 -0.028 -0.007 0.032* -0.005 (615.340) (0.023) (0.018) (0.018) (0.018) Constant 1,977.3** 0.059 0.091*** 0.026 0.095*** (961.908) (0.036) (0.030) (0.029) (0.028) Observations 255 145 291 291 145 R-squared 0.082 0.379 0.257 0.124 0.193 Province FE Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
357. The extent to which use of inputs varies between more and less productive farmers
suggests that a quantile regression may be appropriate. Instead of estimating the
regression relative to the mean, quantile regressions estimate the equation relative to
different percentiles of the conditional distribution of the dependent variable (here
productivity). In particular, a quantile regression can be used to see if different
relationships apply for different parts of the distribution. Quantile regressions for the 10th,
25th, 50th, 75th and 90th percentiles are presented in Table 47. The strong association
between fertilizer and pesticide use and productivity is evident in all of the regressions, but
the magnitudes of the impacts, especially of fertilizer, are particularly large in the higher
percentiles. Given the average productivity by percentiles showed in Table 47, it is possible
to estimate the relative impact of fertilizers.
358. The use of fertilizers increases productivity by a factor of 4.14 in the 10th percentile,
5.62 in the 25th percentile, 4.72 in the 50th, 6.65 in the 75th and 4.13 in the 9th one.
(Table 47.) Fertilizer impacts on productivity everywhere, and the impact seems to be
consistently large. The use of improved seeds though only has a positive impact in the
higher percentiles: in other words their productivity impact is not significant at the bottom
of the distribution. Credit also matters for the most productive farmers, as well as some
less productive ones.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 164
Table 47: Quantile regression for correlates of total productivity in crop cultivation
Q10 Q25 Q50 Q75 Q90
VARIABLES Total Total Total Total Total
seeds -4.3 3.1 32.9 221.1* 740.5**
(8.819) (14.562) (44.495) (134.350) (318.448)
fert 142.0*** 418.7*** 1,474.1*** 4,108.9*** 7,883.3***
(10.821) (17.869) (54.598) (164.857) (390.757)
pest 40.3*** 109.9*** 488.8*** 1,619.0*** 3,583.6***
(9.407) (15.533) (47.463) (143.311) (339.688)
estrume -8.9 -7.7 -49.4 -111.0 30.5
(12.990) (21.450) (65.540) (197.895) (469.066)
credito 2.4 47.0** 58.3 297.9* 1,440.7***
(11.815) (19.510) (59.614) (180.000) (426.652)
precrad 9.1 32.7*** 69.4** 149.7 284.0
(6.515) (10.757) (32.869) (99.246) (235.242)
precass -13.7 -5.8 -21.9 46.0 -73.3
(14.687) (24.251) (74.099) (223.739) (530.325)
precext 11.2 12.2 47.1 133.6 559.4
(12.593) (20.794) (63.535) (191.841) (454.718)
precpub 21.6 53.0** 89.6 262.8 560.0
(14.954) (24.693) (75.451) (227.821) (539.999)
precout 1.8 -0.7 -10.2 -48.4 -108.6
(7.335) (12.112) (37.010) (111.749) (264.877)
Constant 77.7*** 218.7*** 758.9*** 2,161.4*** 5,606.1***
(10.867) (17.944) (54.829) (165.555) (392.412)
Province FE Yes Yes Yes Yes Yes
Year FE Yes Yes Yes Yes Yes
Observations 8,005 8,005 8,005 8,005 8,005
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
6.3. Conclusions on determinants of agricultural productivity
359. There is certainly a strong association between productivity in agriculture and poverty
in Mozambique. Without doubt levels of productivity in agriculture are low and have only
improved very slowly over the 2002-08 period. The use of fertilizer, pesticides, and
improved seeds are all very low; access to credit is limited; and only a minority of farmers
have access to information, and when they do it very rarely comes from extension services.
360. Yet, the picture is not just a negative one. Strikingly, there is a very strong and very
large association between the use of fertilizer and pesticides and productivity. Quite a few
farms grow cash crops and farmers sell quite a high proportion of their output (some 40%
in total). Improved seeds have a positive but more marginal impact on productivity.
361. Data also show that productivity has increased in several locations in Mozambique
over the 2002-08 period, and for the most part these are the same locations which
experienced greater poverty reduction over the same period. These are also mostly the
same locations where fertilizer and pesticides are more widely used. Access to inputs
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 165
seems to be critical to raising productivity and the large majority of farmers fail to get
access to these inputs or to the credit needed to finance their purchase. This would seem to
be a major factor behind their poverty.
362. Access to inputs and credit is clearly not just the responsibility of government, as the
private sector also plays an important role but key government actions can certainly
facilitate this. Infrastructure is one key factor, including access to roads and markets.
Measures to facilitate access to credit in rural areas by simplifying the legal and
institutional framework for credit providers are likely to be useful. Actions to improve
awareness of the importance of fertiliser and other inputs and to incentivise their
marketing and distribution into more remote rural areas might also be considered.
7. Education Sector Outcomes and their Determinants
363. The education sector is crucial in ensuring long-term poverty reduction as well as
sustainable development. It represents a substantial item of public spending - around one
fifth of the public budget and, during the evaluation period, public spending in the
education sector rose dramatically – from some US $ 310 million in 2005 to US $ 938
million in 2012. Moreover, as noted in Chapter 4, it has been the major beneficiary of
Budget Support funding. The analysis of education sector outcomes therefore allows us to
judge whether positive benefits have been obtained from the increased public spending
facilitated by Budget Support.
364. This Chapter is organised as follows. It first assesses the situation in the education
sector in terms of enrolment rates, analysing trends in time, by province and by gender.
Secondly, we present our analysis of the determinants of a selected number of key
outcomes, with particular attention to the schooling inputs (teachers, text books, etc)
provided, directly or indirectly, by government – in other words, the immediate outputs of
education sector spending. This analysis sheds light on the nature of education outcomes,
in terms of quantity and quality, and on the specific expenditure outputs that have been
most relevant in determining such outcomes.
7.1. Overview of enrolment & other key data
365. Data on education is available for Mozambique from two main sources: the Ministry of
Education and the three household surveys carried out in 1996/97 (IAF), 2002/03 (IAF) and
2008/09 (IOF). The Ministry of Education collects administrative data, through the
Education Management Information System (EMIS). The EMIS dataset is a comprehensive
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 166
source of information on pupils and their progress in education, teachers and their
qualifications, schools and their infrastructure. The EMIS data can be disaggregated by
grade, district, school, and gender. It includes information on over 13,000 schools: private
(1%), public (97%) and community (2%) for a period of 9 years (2004 to 2012).
366. The EMIS data on enrolments and other key variables is presented below, It provides
the basis for the econometric analysis of outcomes and their relationship to spending
outputs. In the latter we consider the following outcomes that refer to education quantity
as well as quality:
Absolute enrolment, taking the value of 1 in 2004
Progression indicates the proportion of pupils in grade g that progressed to grade
(g + 1).
Repetition captures the proportion of pupils required to repeat a year before they
can progress further in their education.
Approval indicates the proportion of pupils that were approved in grade 7 to
proceed to secondary education.
367. The econometric analysis relates these outcomes to a number of educational inputs
that can be measured with the EMIS data, such as the ratio of teachers per pupil, the
proportion of teachers with high qualifications, class size, number of classrooms, the
number of mathematics books per pupil, the material used to build the school. All outcome
and input indicators are available at a disaggregated level, for instance by sex, school type,
level of schooling, and by year.
368. The IAF/IOF household surveys are representative at the national and provincial level.
They contain information on education at the individual level, and particularly on children’s
attendance at school and the current grade. This allows us to compute enrolment rates by
province, by gender and by level of schooling. Enrolment rates indicate the share of
children of official schooling age that attend school at a grade corresponding to their age
range – in other words, these are Net not Gross Enrolment Rates71.. Because the surveys
are representative at the provincial level at maximum, it is not possible to match them
with EMIS administrative data at the district or school level.
Enrolment data from household surveys
369. The (net) enrolment rate at the primary level in 2008/09 was 75.7%, markedly higher
71 Secondary school includes technical schools and the age groups are defined to correspond to the
education system of Mozambique. In particular children start primary school at 6 years of age, complete the first cycle at grade 5 when they are 11, and finally complete grades 6 and 7 of primary school by 13 years of age. Secondary school involves 5 grades and it is expected to be completed at
19 years of age.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 167
than the 1996/97 (42.4%) and 2002/2003 (65.9%) rates. These figures confirm the great
effort the government has made in increasing enrolment and the tremendous progress
made in this sector over this period. However the Millennium Development Goal of having
all boys and girls in primary education is still far away and even in 2013 is still unlikely to
be met (Visser 2013). It is plausible that this progress in enrolment involved particularly
poorer households that are most likely excluded from education and most affected by
demand side constraints. Fox et al. (2012) confirm that the largest gains in the observed
increase in enrolment is for poorer households, who also benefited particularly from the
reforms of 2004 – notably, from the abolition of fees and the provision of free textbooks.
370. As far as secondary education is concerned, great progress has been made perhaps
due to the lower starting rates. In 2008/09 17.3% of children of the relevant age group
were in secondary education, compared with a rate of 3.1% in 1996/97 and 6.1% in
2002/03.
Figure 32: Regional pattern in Primary and Secondary Enrolment rates, 2008/09
Source: Author’s calculations from IOF 2008/09.
Figure 33: Primary and Secondary Net Enrolment rates by sex and by year, 1997 - 2009
Source: Author’s calculations from IAF 96/97, IAF 02/03 and IOF 2008/09.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 168
371. In terms of provincial trends, progress was made consistently across provinces both in
primary and secondary enrolment. In all provinces, the increase in primary enrolment rates
was relatively quick between 1996/97 and 2002/03 and more moderate between 2002/03
and 2008/09. At the secondary level the picture is different, with a higher pace of increase
in the second than in the first period. This is probably due to a lag effect: with more pupils
completing primary education more children were actually able to continue to secondary
education, resulting in a higher rate of increase in secondary enrolment in the second
period.
372. The provinces with the lowest primary enrolment rates in 2008/09 were Nampula,
Cabo Delgado, Tete and Niassa; the highest enrolment rates were reported in Maputo City,
Maputo Province, Inhambane and Gaza. The picture is very similar at the secondary level.
The provincial pattern in terms of enrolment rates is highlighted in Figure 32, where
provinces are ordered from north to south on the horizontal axis72 and enrolment rates for
2008/09 are measured on the vertical axis. For primary enrolment, Northern provinces
clearly perform worse than Southern ones. This regional pattern is confirmed at the
secondary level but less markedly so.
373. Girls’ enrolment is always lower than for boys in all years and in all schooling levels.
However, the primary enrolment rate for boys in 2008/9 was only 4% higher than the
corresponding figure for girls, indicating a small and gradually reducing gender gap at this
level of schooling. (Figure 33.)This may be due to the abolition of fees from 2004 and
therefore the softening of financial constraints that may reduce the demand for education
for girls in particular.
374. At secondary level the gender gap is much larger, at 40%, indicating a problem related
to girls staying in school rather than entering the system in the first place. Family planning
issues related to marriage and early pregnancy may also explain this result.
Descriptive analysis based on the EMIS data
Table 48: Key indicators for primary and secondary education in Mozambique, 2004-12
Primary
Enrolment Progression Repetition Class Size
Pupils per Teacher
2004 1.00 0.20 18.47 73.80
2005 1.15 0.84 0.11 19.23 82.65
2006 1.24 0.84 0.06 18.55 86.53
2007 1.33 0.85 0.06 19.25 76.99
2008 1.42 0.85 0.06 19.54 76.46
2009 1.44 0.83 0.07 18.62 71.55
2010 1.46 0.83 0.07 18.27 68.02
72 The numbers in the axis are associated to provinces as follows: 1 Niassa, 2Cabo Delgado, 3 Nampula, 4
Zambezia, 5 Tete, 6 Manica, 7 Sofala, 8 Inhambane, 9 Gaza, 10 Maputo Province, and 11 Maputo City
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 169
2011 1.42 0.81 0.07 18.05 62.32
2012 1.41 0.81 0.07 17.44 60.47
Secondary
Enrolment Progression Repetition Class Size
Pupils per Teacher
2004 1.00 0.17 26.22 29.22
2005 1.20 0.85 0.15 26.81 33.42
2006 1.34 0.82 0.12 27.93 40.38
2007 1.50 0.83 0.11 28.13 39.91
2008 1.60 0.84 0.10 27.78 37.49
2009 1.67 0.83 0.10 27.72 38.11
2010 1.71 0.83 0.11 28.91 40.36
2011 1.70 0.82 0.12 29.25 38.09
2012 1.65 0.80 0.12 27.31 36.94
375. As we do not have population data to compute enrolment rates, the enrolment figures
computed from the EMIS data are absolute figures scaled as an index, such that
2004=1.00. In addition this analysis looks at progression and repetition rates, class sizes
and pupil per teacher ratios. Table 48 presents the national results. The pattern here is one
of fast increasing enrolment. It drops slightly from 2010, probably reflecting the
constraints on system expansion on both supply and demand sides. As already seen, a
particularly large increase is observed at secondary level, although starting from a lower
base.
376. The number of pupils per teacher falls significantly over this period at both primary
and secondary level though the absolute levels remain quite high. The average primary
class size at least seems to be declining marginally over recent years. This reduction in
pupils per teacher is observed in most provinces.
377. At both primary and secondary level the progression rates are in the range 80-85%,
declining a little at primary level in more recent years. Repetition at primary level had
been quite high in 2004, but has declined since and levels are now in the range 6-7%. At
secondary level, repetition rates are higher and increasing since 2009. There are no
significantly different patterns by province.
378. By achieving fast increases in enrolment while maintaining relatively high progression
rates, it has been possible to attain dramatic improvements in gross completion rates,
which increased four-fold at primary level and eight-fold at secondary level between 2004
and 2012. UNESCO has described this as ‘perhaps the most prominent success in the
sector’ (UNESCO, 2011). Given the scale of system expansion, it is not surprising that some
fall in progression rates has occurred, given that a higher proportion of pupils from less
well-resourced households is being admitted. At the secondary level, international data
show that the level of parental education is a major determinant of performance and this
is likely to be an important factor here: many of the pupils enrolling into secondary in the
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 170
evaluation period would have come from households where no adult had secondary
education. Although the drop in progression rates is relatively small (84% to 81% in
primary, 85% to 80% in secondary), the fact that this has been accompanied by a fall in
repetitions indicates that those not progressing are increasingly dropping out of the
education system, which is clearly a concern.
Figure 34: Extent of training of teachers in Primary Education, 2004 - 2012
379. EMIS also reports on the extent to which teachers are trained, the availability of books
and the construction of the schools. This information is summarized for primary schools at
the national level in Figures 34, 35 and 36. There has been a large reduction since 2004 in
the number of teachers without any training: a province level analysis shows this to be the
case everywhere, although the process is slower and less consistent in Zambezia than
anywhere else. At the national level there is an increase in the proportion having a high
degree of training; at the province level though this is not observed everywhere and there
are still hardly any highly trained teachers in Cabo Delgado, Inhambane, Manica and
Zambezia, although the proportions with some training are increasing.
380. The availability of books per pupil at primary level has not kept pace with the growth
of enrolments, declining by 30-35% over the period (compared with a 41% expansion in
enrolments).. In terms of classroom construction there has been a modest increase over
the period in those constructed from concrete or brick. Given the significant expansion in
the number of schools and classrooms this is not an insignificant achievement but a
majority of classrooms are still made of maticado73, mud or other generally substandard
materials.
Figure 35: Index of availability of books per pupil in Primary schools, 2004=1.00
73 Maticado is a mixture of non-durable materials such as mud, straw, and canes.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2004 2005 2006 2007 2008 2009 2010 2011 2012
Other
High
Some
No Formation
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 171
Figure 36: Classroom construction material; Primary Schools, 2004-2012
7.2. Analysis of the determinants of educational outcomes
381. The econometric analysis looks at the four outcomes described above as the dependent
variables. These outcomes are explained with educational inputs, typically provided by the
government, related to teachers, schools and books. The inputs considered are:
Ratio of teacher per pupil
Interaction terms between the ratio teacher/pupil and community/private dummies,
where public schools are the base category
The ratio of highly trained teachers (‘teachers with high formation’)74 to pupils
Interaction terms of the above ratio with community/private schools dummies
(public is the base category)
The number of maths books per pupil
The number of pupils per classroom
The number of classrooms for each school, which indicates school size
The proportion of rooms constructed with bricks or cement
74 This includes foreigners, IMAP, Magistério Primário, UEM-CFP 10th and 11th,UEM-CFP 7 and 9, UP-
Bacharel and UP-Licenciado.
0,00
0,20
0,40
0,60
0,80
1,00
1,20
2004 2005 2006 2007 2008 2009 2010 2011 2012
Portuguese
Math
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2004 2005 2006 2007 2008 2009 2010 2011 2012
Other
Mud
Maticado
Brick
Concrete
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 172
The proportion of rooms constructed with mud
The proportion of rooms constructed with maticado
Participation in the ADE (Apoio Direito as Escolas) program at district level: a direct
support to schools programme targeted to poorer districts, funded through the
FASE programme.
382. Different sets of these explanatory variables are included in the regressions,
differentiating by grade and sex where relevant. Note that the number of Portuguese books
is not included, although data would be available, due to multicollinearity with maths
books.
383. As far as the econometric methods are concerned, the panel nature of the EMIS
dataset allows using a fixed effects estimator that eliminates unobserved factors that
vary across schools but not in time. This includes both school specific elements such as the
ability of the director and the school’s culture and reputation, that are expected to change
little in the relatively short time considered; and location specific effects. Because school
fixed effects also capture location effects, the inclusion of district and provincial dummies
is not required in this context. The fixed effects used in the regressions are taken at the
most disaggregated level and they refer to grade-school combinations. In addition, year
dummies are included in all regressions to take into account particular events that may
have affected the education sector, in a given year.
Table 49: Results for absolute enrolment (1) (2) (3) VARIABLES Total Male Female
Number of classrooms 0.035*** 0.032*** 0.047*** (0.001) (0.001) (0.001) Concrete or Brick 0.006 0.005 -0.000 (0.004) (0.004) (0.008) Maticado 0.006 0.009* -0.036*** (0.005) (0.005) (0.011) Mud 0.017*** 0.017*** 0.006 (0.004) (0.005) (0.010) Constant 0.818*** 0.832*** 0.770*** (0.005) (0.005) (0.011) Observations 351,160 351,045 348,858 R-squared 0.097 0.062 0.074 Number of id 41,317 41,299 41,029 Year FE Yes Yes Yes
Note: standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1
384. The first set of results investigates the absolute enrolment index (2004 = 1.00). Table
49 presents the results for all grades. The explanatory variables here are mainly related to
the school’s infrastructure, with the aim of identifying the inputs at the school level that
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 173
influence enrolment. The number of classrooms has a positive association with enrolment
in all specifications, which is clearly expected. A higher share of rooms built with mud has a
positive association with total enrolment, mainly driven by male enrolment whereas it is
not significant for females. This result may indicate that newer classrooms are built with
non-durable materials and they have therefore a high impact on enrolment. The higher
impact of new classrooms may be in rural areas that typically present lower enrolment
rates and a higher prevalence of non-durable construction materials.
385. Table 50 focuses on progression rates, namely the proportion of pupils that progress to
the following year. The EMIS dataset however does not track individual pupils, but it
provides information about the number of pupils in each class. The underlying assumption
for the calculation of the progression indicator is therefore that pupils do not change
school. We recognize that this is a limitation, since pupils can move school in the course of
their studies but this should not affect the general pattern observed. The progression rate
is constrained to take a maximum value of 1 and it is calculated at the school-class level
to reduce measurement error.
386. In looking at progression, the analysis relates the dependent variable at time t with
independent variables at time (t – 1). The first year of primary education is therefore
excluded. School fixed effects and year dummies are included, and the results in Table 50
include all grades in both primary and secondary education. Because information on books
is only available for primary education, this variable is not included in this set of results.
387. The teacher-to-pupil ratio is important in ensuring progression of pupils in all
specifications, although the effect is smaller for community and private schools. More
highly trained teachers often have a negative influence on progression, though this does
not apply in private schools. Higher training provides more effective teachers, but this may
be outweighed if they apply higher standards to decisions about pupil progression. In
specification (4) we re-run specification 1 with a dummy variable to indicate whether the
district where the school operates is an ADE district75. Results remain the same, but the
rate of highly trained teachers in community schools has now a positive and significant
effect. The positive result for class size in all regressions is rather surprising. An
explanation may be related to the higher pressure to make the pupils progress in big
classes that would become even bigger with a large number of repetitions.
Table 50: Results for Progression Rates (1) (2) (3) (4) VARIABLES Total Male Female Total with ADE
75 Since specifications 1- 3 are fixed effect estimations at school-class level, the introduction of a
dummy that is the same all years (province or ADE district) will be dropped from the estimation. In specification 4 we include province and class dummies in order to control for fixed effects at that level.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 174
Teacher per Pupil 2.82*** 2.66*** 2.30*** 1.81*** (0.06) (0.07) (0.07) (0.04) Teacher per Pupil * Community -2.24*** -2.28*** -1.55*** -0.98*** (0.16) (0.18) (0.19) (0.10) Teacher per Pupil * Private -2.52*** -2.53*** -1.92*** -0.89*** (0.15) (0.18) (0.18) (0.09) Teacher High Education -0.40*** -0.42*** -0.31*** -0.43*** (0.08) (0.09) (0.09) (0.06) Teacher High Education * Community 0.16 0.28 -0.28 0.54*** (0.24) (0.28) (0.30) (0.18) Teacher High Education * Private 0.39** 0.51** 0.31 0.14 (0.19) (0.23) (0.23) (0.14) Class Size 0.01** 0.00 0.02*** 0.15*** (0.00) (0.00) (0.00) (0.00) Number of classrooms 0.00*** 0.00*** 0.00*** 0.01*** (0.00) (0.00) (0.00) (0.00) ADE District 0.00*** (0.00) Constant 0.77*** 0.78*** 0.73*** 0.73*** (0.00) (0.00) (0.00) (0.00) Observations 306,567 306,468 306,390 306,567 R-squared 0.018 0.015 0.009 0.096 Number of id 51,057 51,045 51,039 Year FE Yes Yes Yes Yes School-Class FE Yes Yes Yes No Class FE No No No Yes Province FE No No No Yes
Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1
388. Repetition has slightly different results compared to those of progression. A higher
number of teachers per pupil is negatively associated with repetition in community schools
and for females in public schools, but not in fact in the private schools which few pupils
attend. But consistent with the progression results, more educated teachers are associated
with higher rates of repetition; again the higher level of training may cause them to require
higher standards to allow pupils to progress, especially in years with no national
examination (1-6 grades). If so this may be a positive quality indicator. Repetition rates
are smaller in larger classes (consistent with the same finding in the progression equation)
and where there are more rooms in the school. In this case allowing for ADE districts
makes insignificant the effect of teachers per pupil in public schools, and makes no
difference in the rest of explanatory variables, though the ADE variable itself is significant.
Table 51: Results for Repetition (1) (2) (3) (4) VARIABLES Total Male Female Total with ADE
Teacher per Pupil -0.05* -0.04 -0.07** -0.01 (0.027) (0.030) (0.033) (0.018) Teacher per Pupil * Community -0.32*** -0.33*** -0.31*** -0.44*** (0.064) (0.074) (0.081) (0.041) Teacher per Pupil * Private -0.03 -0.09 0.01 -0.41*** (0.061) (0.076) (0.074) (0.039) Teacher High Education 0.17*** 0.18*** 0.18*** 0.17*** (0.036) (0.040) (0.043) (0.030) Teacher High Education * Community 0.18** 0.17* 0.18 -0.10
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 175
(0.088) (0.101) (0.108) (0.070) Teacher High Education * Private -0.04 0.04 -0.08 0.02 (0.080) (0.097) (0.098) (0.060) Class Size -0.03*** -0.03*** -0.04*** -0.01*** (0.002) (0.002) (0.002) (0.001) Number of classrooms -0.00*** -0.00*** -0.00*** 0.00*** (0.000) (0.000) (0.000) (0.000) ADE District -0.00* (0.000) Constant 0.21*** 0.21*** 0.22*** 0.16*** (0.001) (0.001) (0.001) (0.001) Observations 462,668 462,431 461,272 462,668 R-squared 0.175 0.147 0.133 0.207 Number of id 68,708 68,680 68,637 Year FE Yes Yes Yes Yes School-Class FE Yes Yes Yes No Class FE No No No Yes Province FE No No No Yes
Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1
389. The key regressions of interest here are those that look at the rate of pupils approved
to proceed after grade 7 of primary school, i.e. those who are allowed to enter secondary
school. The results only refer to pupils in grade 7 and the fixed effects are included at the
school level. It is useful to report that approval rates are very high in Mozambique (80 %
for boys and 79 % for girls.) Therefore the marginal effect of the independent variables
can be expected to be relatively small.
390. Table 52 shows that the number of teachers per pupil has a positive effect on the
outcome of interest for the case of females only or in the specification including the ADE
district dummy. A higher teacher to pupil ratio helps in following pupils more closely and
therefore facilitating their approval to proceed to the next level of education. The same
variable, however, has a negative effect in community schools. But the most striking result
in this model is that the higher education of teachers seems to have a negative effect on
approval, and this is significant and quantitatively important in all models. The result may
be explained by higher standards under highly qualified teacher that may tend to be
stricter with pupils therefore making it more difficult for them to be approved at the end of
grade 776. Class size has the expected negative and significant coefficient in most
specifications, consistent with the result on teacher-to-pupil ratio. Since this is a primary
education model, we could include the books (maths) per pupil variable. This variable has a
positive association with the outcome for males and in the model controlling for ADE
districts. In addition, approval rates are lower in ADE districts.
Table 52: Results for approval into Secondary School (1) (2) (3) (4) VARIABLES Total Male Female Total with ADE
76 There could be other factors. Higher qualified teachers will work in the more urban schools, where also
school inspection would be more a reality than remote rural schools
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 176
Teacher per Pupil 0.62** 0.43 1.12*** 0.62*** (0.245) (0.275) (0.337) (0.161) Teacher per Pupil * Community -2.19*** -0.40 -2.90*** -0.45* (0.615) (0.902) (0.828) (0.268) Teacher per Pupil * Private 0.20 0.03 0.43 0.70* (0.967) (1.088) (1.300) (0.382) Teacher High Education -0.69** -0.56* -1.31*** -0.64*** (0.274) (0.308) (0.380) (0.220) Teacher High Education * Community -0.35 -4.42*** 0.92 1.01** (0.908) (1.282) (1.222) (0.476) Teacher High Education * Private 0.80 0.70 0.84 0.35 (1.101) (1.238) (1.480) (0.573) Math Books 0.01 0.01* 0.00 0.02*** (0.005) (0.006) (0.007) (0.004) Class Size -0.05*** -0.05*** -0.04 -0.13*** (0.017) (0.019) (0.023) (0.010) Number of classrooms -0.00 -0.00* -0.00 -0.00*** (0.001) (0.001) (0.001) (0.000) ADE District -0.01** (0.003) Constant 0.86*** 0.87*** 0.84*** 0.92*** (0.010) (0.011) (0.013) (0.007) Observations 10,035 10,026 9,979 10,035 R-squared 0.110 0.087 0.074 0.131 Number of id 3,384 3,381 3,375 Year FE Yes Yes Yes Yes School-Class FE Yes Yes Yes No Class FE No No No Yes Province FE No No No Yes
Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1
7.3. Conclusions: key determinants of education outcomes
391. This report presents an analysis of the education sector based on different data
sources, namely EMIS and national household budget surveys, and methods, including
descriptive and econometric analysis. In terms of progress over the evaluation period, the
results point to great improvements in education particularly related to the increase of
enrolment rates for both boys and girls.
392. As far as educational inputs are concerned, the econometric analysis pointed to the
importance of the teacher-to-pupil ratio for both approval in 7th grade and progression. In
this respect, it shows clearly that the increased public spending on teacher recruitment and
salaries, as well as on classroom and school construction has been a determinant factor in
the outcomes recorded.
393. The increase in the level of training of teachers is a welcome development, though this
also seems to be associated with higher repetition rates and lower approval in grade 7.
This may be due to higher standards applied by qualified teachers or to inadequate
training, rather than to the limited role of higher qualifications in improving outcomes in
education. It does not suggest to us that it is an inappropriate investment, rather that it
should be supported by additional inputs targeted on those schools with higher repetition
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 177
rates and lower approval in grade 7.
394. Smaller class sizes are positively associated with approval into secondary school, but
are also positively associated with repetition, perhaps because there is a tendency to ‘avoid
repetition at all costs’ in over-crowded schools with large class sizes. Again, it suggests a
need for a cluster of inputs targeted to the schools with the larger class sizes.
395. The availability of maths books is significantly associated with higher approval rates in
some specifications. This would suggest that there are only certain contexts (in terms of
pupil-teacher ratios, levels of training of teachers, etc.) in which increased availability of
text books is likely to make a positive difference. Further analysis could assist in
identifying “threshold levels” (minimum requirements) for these essential complementary
inputs.
396. Despite this positive progress achieved throughout the sector, much remains to be
done to achieve the MDG of having all boys and girls in primary education. Moreover,
enrolment rates in secondary education remain low, despite massive increases in the last
15 years, and the widening gender gap at this level is particularly worrying. All this points
to the need to keep a high priority on the education sector and to continue the efforts
carried out in the evaluation period.
397. On the positive side, our evidence suggests that the combination of educational inputs
on which public spending has been concentrated, notably teachers, teacher training and
classroom construction is efficient and effective. Moreover, these inputs have been
allocated relatively efficiently both geographically and by gender. In some cases, notably
for the schools with the larger classroom sizes, it would seem that more attention is
needed to provide the right combination of additional inputs. Further econometric analysis,
supported by qualitative work in schools may assist in determining the precise parameters
of these optimal combinations and in understanding how they may vary by type of school
and geographical location.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 178
8. Conclusions and Recommendations
8.1. Overall findings
398. This study has evaluated all of the Budget Support operations undertaken in
Mozambique from 2005 to 2012. These operations amount to a resource transfer of US $
3,354 million - an annual average disbursement in excess of US $ 400 million, some $16
per annum per head of the Mozambican population, provided by 19 Development Partners
(the “G-19” or Programme Aid Partners”). In evaluating these operations, our study has
addressed three questions:
i. How successful has Budget Support been in providing the means to the
Government of Mozambique to implement its national and sectoral strategies?
(EQ. 1.1-2.3)
ii. How successful has Budget Support been in facilitating improvements in the
efficiency and effectiveness of these national and sectoral strategies? (EQ. 3.1
– 3.4)
iii. As a consequence, how successful has Budget Support been in attaining
successful outcomes and impacts on growth and poverty reduction? (EQ. 4.1 &
4.2.)
Providing the means to implement Government policy (EQ.1.1 - 2.3)
399. In relation to financial inputs, Budget Support inputs have been important and
efficiently delivered. Disbursements during the 8-year evaluation period were substantial
both in fiscal terms – where they represented on average 15 % of public spending, and as
a proportion of total ODA, where they comprised an average of 30 %.
400. Annual predictability of budget support has been significantly better compared with
the two other aid modalities. Actual disbursements of Budget Support have generally been
close to 100% of the planned disbursements, averaging 116% over the 8-year period since
2004. This compares with projects, which averaged disbursements of 56% of projections
with a considerable variance around this mean. Basket fund disbursements were more
stable, but also averaged significantly below Budget Support as a percentage of
projections. (79% versus 116%),
401. The objectives of in-year predictability and of “front-loading” of disbursements were
largely fulfilled during the evaluation period. Budget Support providers were clearly
conscious of the importance of these objectives and managed to maintain good levels of
predictability even through a period of difficult financial circumstances in many donor
countries.. Over 2008 – 2012, an average 60-80 % of BS funds were disbursed in the first
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 179
two quarters (i.e”front-loaded”). From 2005 to 2009, the gap between quarterly projections
and actual disbursements was close to zero. In-year predictability was compromised in
2010, 2011 and 2012 due to suspensions resulting from concerns over potential violations
of underlying principles, although the financial costs of these temporary suspensions were
not substantial.
402. Budget Support is no longer exerting a significant influence on the overall
effectiveness of aid. In the early years of the evaluation period and in the years
immediately preceding it, the Budget Support process was instrumental in generating
support for aid effectiveness principles. A key aspect of this was the commitment by the
Programme Aid Partners to monitor their own performance through the PAPs’ Performance
Assessment Framework. This was widely heralded as major step towards mutual
accountability and highly valued by the Government of Mozambique. However, since 2009,
performance against these indicators has plummeted. A firm renewal of the commitment
to these objectives and a re-investment by the Government of Mozambique and its
Development Partners in the aid effectiveness principles, will be required for performance
to improve again.
403. Regarding the use of country systems, the importance of Budget Support within total
ODA has decreased during the evaluation period (from 30% to 20%). Moreover, there is no
evidence of a lasting influence of Budget Support on other modalities concerning the use of
the national budget and/or the single treasury account (CUT). The proportion of projects
that were executed off-CUT averaged 85% during the period 2009-2012. Common Basket
Funds have made more use of country systems but there is no evidence that this has been
influenced by Budget Support.
404. By 2005, Mozambique had established a structured, harmonised assessment and
disbursement process for Budget Support, linked directly to the achievement of the targets
in the national poverty reduction strategy. This system - centred around two Performance
Assessment Frameworks (PAFs) one for Government and one for the Programme Aid
Partners (PAPs) - has been preserved. An integrated annual review calendar has been
established, clearly linked to a network of sectoral and thematic policy processes. The
proceedings of the annual review process are open to a wide range of stakeholders, their
results are published and are the subject of press releases.
405. Since 2011, the transaction costs of the dialogue process have increased largely due
to changes to the framework introduced to fit with the newly issued PARP, which adopted a
more cross-cutting, thematic structure. Whilst the rationale for the new dialogue structure
is clear, its introduction has met with “teething problems”: some simplification is needed, in
addition to further guidance on how it is expected to work.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 180
406. The Government PAF fulfils its intended purpose reasonably effectively and compares
well with the PAFs used in other Budget Support recipient countries. It provides a relatively
strategic set of measureable targets, which have worked as a focus for dialogue. The
number of targets is broadly appropriate and the process of assessment does not entail
excessive transaction costs.
407. However, the PAF has inherent limitations, and requires complementary frameworks in
order to work effectively: these complementary dialogue frameworks are in need of further
development. Like all indicator-based assessments of performance, the PAF has three
disadvantages: i) it focuses exclusively on what is measureable, which is not always what
is important; ii) it runs a continuous risk of generating perverse incentives by converting
performance indicators into policy targets, to which money and prestige is attached; and iii)
it relies on pre-existing processes of research and policy development in order to generate
meaningful policy targets. Given these inherent limitations, the PAF should therefore be
neither the sole method of assessment nor the sole locus of dialogue. In Mozambique,
alternative frameworks have been utilised on an ad hoc basis – such as the matrices of
policy actions developed as ‘areas for special attention’. These have presented a relevant,
pragmatic response to the problem but do not constitute a sustainable longer term
solution.
408. Our assessment of the policy dialogue process suggests that the main structural
weakness lies precisely within the policy development process, which should be situated at
the sector and thematic level. There are several problems of strategic importance to the
PARP agenda – such as low productivity in agriculture, high levels of malnutrition, a poor
enabling environment for doing business - where effective policy responses have not been
generated. GBS has had only a limited influence on policy development processes at this
level, for four principal reasons:
The underlying capacity constraints from which Mozambique suffers mean that
policy analysis and development skills have been at a premium.
Sector level processes have their own history and their own dynamics, on which
the influence of the central level GBS PAF has proven to be modest. Where the
sector dynamics have been good, such as in the case of education, the GBS
dialogue has helped to sharpen the strategic focus and to raise awareness of
sector level issues; where the sector dynamics have been less favourable, such as
in agriculture, the GBS process has proven incapable of exerting a positive
influence.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 181
The disinvestment in the Budget Support process, reported to have afflicted both
the DPs and Government, has reduced the administrative attention and analytical
effort, which might be harnessed to address sector level policy problems.
DP stakeholders asserted that there had been a decline in the quality of
coordination within the G-19, and a consequent decline in the ability of the G-19 to
develop well-researched and coherently defined policy positions and to harness
analytical resources for the common good.
409. The inputs provided by the four common funds supporting PFM & Governance
functions, alongside funding lines linked more directly to Budget Support operations,
amounted to a significant TA/ capacity-building input within the Budget Support package.
Due to the multitude of these arrangements and the range of funding lines through which
they have been provided, it has not been possible to reach a judgement about their
efficiency and effectiveness. However, Government stakeholders were positive in their
appreciation of these inputs, both with regard to their relevance and the relative efficiency
with which they were provided.
410. Moreover, there is evidence that the TA/ Capacity-building provided by these common
funds generated important outputs, which facilitated the production of the “induced
outputs” targeted by Budget Support. This has been most notable with regard to PFM
reforms and the strengthening of the capacities of the Autoridade Tributária de
Moçambique (ATM), the Tribunal Administrativo and the Inspeção Geral de Finanças.
Facilitating improvements in Government policies (EQ 3.1-3.5)
411. During the evaluation period, there has been a strong performance by the Mozambican
authorities in fiscal and macroeconomic management:
Revenue mobilisation has increased by nearly 10 points of GDP, out-performing
Tanzania and Uganda and narrowing the gap with Kenya.
Total spending has increased by more than 9 points of GDP as a consequence of
both expanded recurrent and capital expenditure.
Domestically financed development expenditure has almost doubled from 3.2 % of
GDP in 2005 to 6.1 % in 2012.
The overall deficit before grants increased to nearly 15 % of GDP in 2009 due to
the global financial crisis, but fell to less than 10 % of GDP in 2012, a more
sustainable level.
Although inflation fluctuated at 10-12 % up to 2011, restrictive monetary policy
brought it to 2.2 per cent in 2012, the lowest in the SADC region.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 182
412. Budget Support funding contributed significantly to these achievements, without
introducing substantive negative side effects. It is unlikely that the growth of domestically
financed capital expenditure could have been achieved in the absence of budget support,
without recourse to unsustainable levels of borrowing. However, macroeconomic and fiscal
policies have been driven by the Government’s own objectives and its PSI agreements with
the IMF. There is no evidence that the Budget Support process had more than a supportive
effect in this area.
413. Stakeholders report that monitoring progress on the EITI initiative through the PAF
helped to add momentum to the process and thus contributed to the country achieving
compliance. There has been a significant amount of work undertaken by the G19 task force
on Extractive Industries, which was used for political dialogue with the Government. There
was also a PAF indicator from 2010 onwards, monitoring progress on the EITI initiative.
414. There is a risk on the horizon concerning the expansion of commercial external
borrowing for capital investment, taking advantage of the favourable long term outlook
due to revenues from the extractive industry. High commercial interest rates and
demanding principal repayment schedules may result in debt obligations being “front-
loaded” relative to revenue flows. Public investment projects also need to be planned
carefully to ensure that they meet policy priorities and are executed efficiently.
415. Significant improvements have been made in Public Finance Management:
A coherent programme of PFM reform has been in place throughout the evaluation
period, to which the Government has shown a clear commitment;
The GoM has achieved important progress in setting up a modern PFM system
based upon a functioning IFMIS system.
The PEFA assessments of 2006, 2008 and 2010 attest to improvements across all
aspects of the PFM system and in particular with regard to comprehensiveness &
transparency, revenue collection and management, cash management and internal
and external audit.
416. Without doubt, the central place of PFM in the GBS dialogue has had a positive impact
on the pace and content of reform. Also important have been the technical assistance and
direct financial support provided through the 4 common funds.
417. Nevertheless, there are major challenges ahead and it is not clear that an adequate
framework exists to address these challenges:
The 2011 – 2015 PFM Vision presents an ambitious reform programme but
exhibits weaknesses regarding the sequencing of reform and the scope of reforms:
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 183
more attention should be dedicated to the monitoring of fiscal risks, within a
“whole of government” perspective.
In relation to the coordination of reforms, problems remain over the adoption of
CEDSIF as the reform coordinator: the coordinator of reforms should ideally have a
wider perspective on PFM issues and a higher status, in order to exert authority.
Regarding the advice and support available through dialogue with the G-19, there
is a question mark over the ability to marshal adequate technical expertise to give
guidance on reform sequencing and other increasingly technical issues now
emerging.
418. Total spending on the priority sectors designated in PARPA & PARP has more than
quadrupled in nominal Meticais terms over the evaluation period, increasing by slightly
more than 7 percentage points of GDP. As a percentage of total expenditure, these priority
sectors have increased their share from 61 % to just over 67 % of total spending.
419. The major contribution of GBS funding has been to support the expansion of public
spending in the education and good governance sectors, alongside a lesser contribution to
the agriculture sector. We draw this conclusion from an analysis of trends in funding
sources and sector expenditure shares, making assumptions that reflect the processes of
budgetary negotiation and decision-making in Mozambique.
420. There is strong evidence that the combination of GBS funding through the budget with
targeted support to institutions of accountability such as the NAO, the Parliament, CSOs
and the media has created more transparency. Targeted support through common funds
has been fundamental to this achievement but Good Governance is one of two sectors that
appear as the primary beneficiaries of Budget Support funds.
421. Although there remain concerns about the quality of transparency and accountability in
Mozambique, over the period of the evaluation there is evidence of improvement:
The public availability of budgetary documents improved as measured by its score
on the Open Budget Index (OBI) which increased from 28 % in 2010 to 47 % in
2012.
There has been a sharp improvement in the number and the quality of audit
reports of the Tribunal Administrativo from 29 audits in 2004 to over 400 per year
since 2009.
CSOs and the media have become more active on corruption issues and in
reporting on budgetary and financial management questions.
422. The legal and institutional framework for combating corruption has been strengthened
during the evaluation period:
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 184
In 2004, Parliament approved Mozambique’s first Anti-corruption Law (Law nº
6/2004).
In 2005 the Public Procurement Law was passed and then subsequently refined in
2010 through Decree 15/2010.
As part of the Governance Action Plan, a package of 5 anti-corruption laws was
submitted to Parliament during 2010, with 3 of these 5 laws being passed in 2012.
Measures have been taken, since 2010, to strengthen the budget and the staffing
of the Central Office for Combating Corrupting – Gabinete Central de Combate a
Corrupção (GCCC).
423. However, there are concerns over the extent to which anti-corruption legislation is
being implemented. A higher number of corruption cases has been taken up by the GCCC
and a bigger proportion of these have been taken to court but there are widespread
concerns over the politicisation of the judicial process. As a result, there is little confidence
among CSOs or the general public that the selection of cases for investigation and
prosecution is genuinely impartial and evidence-driven.
424. There is evidence that the high level debate between Government and the G-19 at the
time of the “crispação” was influential in focusing attention and speeding up the
implementation of actions included in the Governance Action Plan (GAP). Government
stakeholders asserted that all of the issues included within the GAP were already under
internal discussion and would have been implemented in time but they agreed with DPs
that the high-level attention to these issues helped to speed up their implementation.
Attaining sustainable outcomes and impacts (EQ. 4.1 and 4.2)
425. GDP growth remained high throughout the evaluation period, averaging 7.3 per cent
annual growth. Moreover, the pattern of growth remained stable throughout the period
despite the global financial crisis.
426. Budget Support funds have enabled the government to increase development spending
without recourse to domestic borrowing. This has helped to keep gross domestic
investment buoyant (at 19-20 % of GDP) up until 2011 when large FDI inflows linked to
the extractive industry then kicked in. Budget Support facilitated an expansion in
domestically financed development spending from 3.2 % of GDP in 2005 to 6.1 % in 2012.
By avoiding recourse to borrowing, the Government also restrained its demands on the
domestic banking sector and facilitated the increase in the ratio of private sector credit to
GDP.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 185
427. Mozambique has made efforts to improve the investment climate but still ranks low on
international indicators of economic competitiveness and the business environment.
Dialogue in this area has largely failed to achieve any acceleration in progress.
428. There has been no statistically significant change in poverty between the early and the
late 2000s, with 54.1% of the population reported below the poverty line in 2003 and
54.7% in 2009, according to the official poverty figures, presented in the Third National
Poverty Assessment. Therefore, there seem to be a missing link between growth and
income poverty reduction. While the country experienced high and sustained growth rates,
the poverty reduction performance was disappointing. Indeed the analysis of pro-poor
growth shows that at the national level the poor have benefited less from growth than the
non-poor.
429. In part, this may be attributed to the pattern of public spending, which had largely
favoured the education and heath sectors – investments which might support poverty
reduction in the long term but which would not have a positive short-term effect on
household consumption. In part, it may be attributed to the slow progress in improving the
business environment, which would have hindered the growth of small and medium-scale
enterprises; and, in part, to the failure to generate improvements in the productivity of
small and medium farmers in the agriculture sector.
430. The sectoral analysis of GDP shows that many sectors experienced good growth,
including agriculture. This sector is particularly important in the context of a highly rural
population (estimated at 70%) and given its large contribution to GDP (39 % in 2011).
However agricultural growth, that was faster in the evaluation period than before, failed to
benefit the poor.
431. There is a very strong and very large association between the use of fertilizer and
pesticides and productivity. Access to inputs seems to be critical to raising productivity and
the large majority of farmers fail to get access to these inputs or to the credit needed to
finance their purchase. This would seem to be a major factor behind their poverty.
432. GBS dialogue had little impact on the development of policies to improve access to
inputs and credit within the evaluation period, despite the fact that agriculture and rural
development were prominent in the PAF. There are, however, indications that the increased
level of public spending in the sector since 2010 has in part been facilitated by GBS.
433. Non-income measures of poverty have shown some improvements:
From 2005 to 2012, Mozambique’s Human Development Index increased from
0.287 to 0.327. (HDR, 2013).
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 186
Indicators of malnutrition have improved moderately, while still remaining very
high by international and regional standards.
The 2008/09 Household Budget Survey showed improvements in some indications
of welfare, and most especially in education provision.
434. Education was designated as a ‘priority sector’ in PARPA/ PARP and has throughout the
evaluation period commanded the highest share of the national budget. Combined with
some signficant policy changes introduced in 2004 (such as abolition of school fees,), this
has permitted dramatic improvements in education provision:
The net enrolment rate at the primary level in 2008/09 was 75.7%, markedly
higher than the corresponding rates in 1996/97 (42.4%) and 2002/2003 (65.9%).
41 % more children were in school in 2012 in comparison with 2004.
The largest gains in the observed increase in enrolment were for poorer
households. (Fox et al., 2012)
17.3% of children of the relevant age group were in secondary education in
2008/09, compared with a rate of 3.1% in 1996/97 and 6.1% in 2002/03. Pupil
numbers were 65 % higher in 2012 as compared with 2004.
In terms of provincial trends, progress was made across all provinces both in
primary and secondary enrolment.
The primary enrolment rate for boys in 2008/9 was only 4% higher than for girls,
indicating a small and gradually reducing gender gap at this level of schooling.
At secondary level the gender gap is much larger, at 40%, indicating a problem
related to girls staying in school rather than entering the system in the first place.
435. Econometric analysis pointed to the importance of the teacher-to-pupil ratio for both
approval in 7th grade and progression. In this respect, it shows clearly that the increased
public spending on teacher recruitment and salaries, as well as on classroom and school
construction has been a determinant factor in the outcomes recorded. Moreover, evidence
suggests that these inputs have been allocated relatively equitably geographically and by
gender.
436. Funding through Budget Support has played a key role in funding the expansion of the
education sector. The expanded levels of provision could not have been financed in the
absence of General Budget Support, which facilitated the expansion of the teachers’ wage
bill and of internally financed investment in school construction, while FASE common
basket funding and projects were dedicated to school construction, teacher training and
supporting inputs. The policy framework for the fast expansion of public education was
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 187
developed primarily through FASE structures, although the GBS framework helped to
generate support at the centre of government, most notably from the Ministry of Finance.
8.2. Step Three: Conclusions on Impact of Budget Support
437. What do the findings from Step One and Step Two tell us about the overall impact of
Budget support over the evaluation period? In order to address this question, we have
constructed a matrix (Figure 37) with the following elements:
It presents on the right-hand side the targeted outcomes and impacts - growth,
reductions in income and non-income poverty and improved education sector
provision.
In the middle column, it identifies the changes in public policies and expenditures,
induced by Budget Support, which might have exerted a positive influence on final
outcomes and impacts.
In the left-hand column, it identifies the inputs and immediate effects of Budget
Support.
In the two connecting columns it identifies the relative degree of influence, which
the evaluation team have identified, defining this as (i) Absent or Negligible; (ii)
Weak; (iii) Moderately Important; and (iv) Important.
438. A simplified diagrammatic overview of our Step Three analysis is presented in Figure
38. This should be read together with the matrix in Figure 37 so as to understand the lines
of causality identified77.
77 I t should be noted that this is diagram aims to capture effects of changes within the evaluation period.
Hence, the fact that major gains in the use of country systems were made over the 2000-2005 period – largely as a consequence of Budget Support – is not reflected here. Those gains would have had a lasting impact on the quality of resource management by the Government of Mozambique but within the evaluation period itself, no additional incremental gains in use of country systems were identified.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 188
Figure 37: Step Three – Summary of analysis of aggregate impact of Budget Support in Mozambique
BUDGET SUPPORT INPUTS &
IMMEDIATE EFFECTS
NATURE & DEGREE OF
INFLUENCE
GOVERNMENT POLICY CHANGES WITH
POTENTIAL INFLUENCE ON OUTCOMES
NATURE & DEGREE OF
INFLUENCE
KEY IMPACTS & OUTCOMES
IDENTIFIED
Budget Support Funds,
predictably delivered, for
relatively low transaction costs
Increased use of Budgetary
process by Aid as a whole
Increased Harmonisation &
Alignment of Aid
Framework for Policy
Monitoring & Dialogue
Capacity Building
Important – disbursements of
av. 15 % of public spending
have helped fund
development budget & some
recurrent spending, while
supporting a sustainable
deficit.
Negligible – use of country
systems has not increased
substantially in the evaluation
period, and relative weight of
Budget Support has declined.
Negligible – in recent years,
there has been reduced
attention to H&A by DPs.
Moderate – the PAF system
for target setting, monitoring
& review is in place and
working relatively effectively
together with other
complementary channels of
dialogue.
Moderate – There was
substantial TA support to PFM
Stable Macroeconomic Management &
Growth-oriented fiscal policy
Enhanced allocations to PARPA II / PARP
priority sectors
Strengthened PFM system, supporting
more efficient expenditure
Improved Sector Policies
Improved Democratic Accountability
Important – Good macro
performance preserved
confidence & investment
supported Growth.
Important – Allocations to
priority sectors rose by more
than 7 % of GDP
Negligible – PFM system
improvements still too weak
for major efficiency effects
Moderate – policies on
Education were effective but
agricultural policy not focused
on small farmer productivity,
and limited progress on
business environment.
Weak – Transparency has
improved but with limited
effects on accountability.
Parliamentary oversight is
still weak and Anti-Corruption
7.3 per cent GDP growth per
annum
No change in income poverty
– from 54.1% in 2003 to
54.7% reported below
poverty line in 2009
Modest reductions in non-
income poverty, but
malnutrition still a major
issue.
Major Improvements in
coverage of Primary and
Secondary Education, while
approaching gender parity at
primary level.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 189
& Governance linked with
GBS.
bodies are not yet fully
utilising their legislative
powers.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 190
Figure 38: Diagrammatic Overview of Budget Support impacts and transmission mechanisms in Mozambique
1. Budget support
inputs Strength of linkage 2. Induced outputs Strength of linkage 3. Impacts
Strengthoflinkage:
Important
Moderate
Weak
Absent
1.1 Funds
1.3 Dialogue
1.4 Capacity
building 2.5 Improved
Policies
2.4 Increased
priority sector
spending
2.2 Stable macro
management
2.1 Increased Public
Investment 3.1 Economic
growth
3.2 Reduction in
income poverty
3.3 Reduction
in non-income
poverty, inc.
expanded
Education
coverage 2.6 Improved
democratic
accountability
2.3 Improved PFM
1.2. Incremental
use of Govt
systems
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 191
Summary of Step Three Analysis: The Impact of Budget Support in Mozambique
439. The primary contribution of Budget support has been in the form of increased funding.
This was provided in a manner which was both more predictable and lower in transaction
costs than other modalities.
440. The potential wider effects of Budget Support on aid effectiveness have not been
evident within the evaluation period, although it is possible that use of country systems
and harmonisation and alignment would have been lower in its absence. Our judgement is
that these effects were strong in the years shortly before the evaluation period but that
there has been no further increment in the use of government systems during the period
evaluated.
441. Budget Support inputs for technical assistance and capacity building have been
significant in scale. These were provided primarily through common basket funds linked to
PFM reform and to the strengthing of the governance institutions. We judge that these
inputs had a moderately important influence on improved PFM processes and a weak
influence on improved democratic accountability.
442. Budget Support has supported the creation of an effective structure for dialogue,
based upon the definition of policy targets and a framework of annual monitoring,
comprising sector reviews and a national level policy dialogue, as well as periodic
assessments of underlying principles. Nevertheless, the influence of this framework has
been limited by underlying weaknesses in sectoral processes of policy analysis and
development – a weakness which GBS processes could only mitigate to a modest extent.
The fact that PFM reform and improvements in governance, especially the fight against
corruption, were systematically discussed and reviewed on an annual basis through the
Budget Support dialogue is judged to have been moderately important to the progress
made in PFM, and also a source of influence upon the more modest progress made in
democratic accounatbility.
443. Through its funding contribution, Budget Support had an important influence on
macroeconomic management and on the increased allocation of resources to PARPA/ PARP
priority sectors. Budget Support funds are judged to have been important in allowing high
levels of public investment to be maintained, and by providing concessional and grant
funding for such investment, potential negative “crowding out” effects on private
investment were also avoided.
444. Good macroeconomic management, combined with increased allocations to the priority
sectors allowed Budget Support to have an important influence on overall growth and on
the dramatically improved outcomes in the education sector. Even if the desired reductions
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 192
in poverty did not occur, these are major achievements, which fully justify the provision of
Budget Support.
Could other modalities support scaling up of service delivery as effectively as Budget Support?
445. Neither project funding nor common basket funding could have achieved the same
results as Budget Support, with the same degree of efficiency, effectiveness and
sustainability:
The funding provided through GBS of US $ 450 million per annum would have been
equivalent to about 100 additional project and CBF operations, disbursing an
average of $4-5 million each per annum. These additional transaction costs would
almost certainly have been prohibitive, given existing constraints on absorptive
capacity.
It would have been difficult for GoM to utilise such operations to finance
recruitment of additional teachers or to provide non-salary recurrent cost funding.
With the average levels of predictability of disbursements from projects (56%) and
of CBF operations (79%) achieved during the evaluation period, there would need
to have been a systematic over-programming of some 40 %, and even with such
adjustments the annual volatility of disbursements would have proven
problematic.
The sustainability of this incremental funding would have been difficult to ensure,
whereas Budget Support funding is gradually being replaced with domestic
revenue, using the same planning and budgeting procedures and hence ensuring
sustainability.
Finally, while we have concluded that the influence of Budget Support policy
dialogue has been relatively modest, it has served to generate certain
improvements in the design and implementation of policies and strategies, which
could not have been generated by the more limited spaces for dialogue offered by
project and CBF arrangements.
446. The analysis of common basket funding (CBF) mechanisms has not formed a major
part of the terms of reference for this evaluation but the prominence of the PROAGRI, FASE
and PROSAUDE mechanisms in agriculture, education and health has permitted at least
some analysis of these three mechanisms. The contrasting performance of the three is
quite revealing.
447. A central tenet of good public finance is that there should always exist an effective
“budget challenge function”, which is normally provided by the Ministry of Finance and/or
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 193
the Ministry of Planning, as well as by the Legislature. This “budget challenge” requires all
sectoral proposals for spending to be assessed against competing demands from other
sectors, against national policy priorities and against the prevailing rules of financial and
personnel management. In the absence of this challenge from the centre, sectoral
agencies will tend to assign themselves larger budgets than necessary and to spend them
less on service delivery functions and more on sectoral administration functions and
processes, from which they directly benefit.
448. A common weakness in many CBF mechanisms is that they inadvertently by-pass the
budget challenge function by excluding both the Legislature and the Ministry of Finance
and other central agencies from the process of scrutinising proposals for spending from
the sectoral basket fund. In many CBFs, sector ministries submit proposals directly to
sector donors, without the involvement of the Ministry of Finance. Sector donors may
sometimes be able to apply an adequate budget challenge function but there are often
conflicts of interest which prevent this: notably, sector donors share the objective of sector
ministries to increase the sectoral budget and will tend to put a lower valuation on the
budget demands of other sectors.
Box 3: Why PROSAUDE is not SBS and how GBS/ SBS could be diferent
In the last decade health sector fragmentation has been aggravated in Mozambique by the
presence of multiple donors and the onset of vertical funds, which now dominate in the sector.
Sector stakeholders responded to this fragmentation by developing a sector SWAp in the 1990s
followed by the establishment of three common funds and then in 2009 their amalgamation into one –
PROSAUDE II. The intention of the health partners involved was to transition to SBS, but in spite of
this, PROSAUDE II has not moved beyond the original arrangement.
PROSAUDE II is on-budget, and on-Treasury, using a dedicated account in the CUT, but it is a
project account, effectively controlled by MISAU and overseen by the PROSAUDE partners. The
funds are not budgeted through standard government procedures and hence not subject to the budget
contest, which that process incorporates. There is no procedural manual, although a broadly based
budget is drawn up for PROSAUDE II spending. It is not controlled or monitored according to
government procedures, and there are no written rules governing budget revisions, virements etc.
PROSAUDE II funds are often spent on items which cannot be covered by the OE such as salary top-
ups, salaries of new employees pending registration on e-CAF, per diems and other personnel items,
which would normally be met by GoM. In this respect they circumvent, rather than follow, government
procedures. This is done with good intent, and in recognition of the serious shortage of human
resources in the health sector. Nonetheless, the process has created a perverse incentive, removing the
urgency to reform health sector salary scales and bureaucratic recruitment procedures.
PROSAUDE II has not made the transition to SBS for several reasons: first because there has not
been a consensus amongst PROSAUDE signatories, in favour of SBS (partly a result of the
documented and persistent limitations of GoM financial systems, and the difficulty in obtaining reliable
audits); second, with the growth of vertical funds, PROSAUDE II financial leverage became less
(averaging around 20% of the Health Budget in recent years but a lower proportion of total resources
for health); and third, MISAU does not favour SBS because it values the flexibility of using
PROSAUDE II funds to cover expenses that cannot be paid from OE.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 194
GBS would be different because funds would flow through the OE and would not be dedicated to
any particular sector or purpose. National accounting systems, including Treasury budget monitoring
systems, would apply, and there would be no opportunity for MISAU to circumvent regulations, for
instance in hiring staff without TA approval. Flexibility would in that sense, be reduced. How would
SBS be different? Funds would flow to an account controlled by MoF (not MISAU) and would be used
to enhance the OE ceilings of beneficiary health agencies. They would be budgeted in concert with OE
funds. Like GBS, SBS funds would form part of a single budget governed according to standard budget
monitoring, review and audit arrangements.
The outcome of both SBS and GBS would be lower transaction costs, greater transparency, and
the ending of the perverse incentives described above.
449. In Mozambique, the PROAGRI and PROSAUDE basket funds clearly suffered from the
lack of a budget challenge. As a result, in both cases a disproportionate share of the CBF
was dedicated, not to service delivery functions, but to expenditures on sectoral
administration processes at the headquarters level, including workshop processes and
institutional development expenditures. In both these cases, CBF funds were also used to
finance staff salaries, often in contravention of public service rules, and without serious
examination of the sustainability of financing salaries from external funds. In the case of
PROSAUDE, over 5,000 heath staff are financed through the common fund, although
initiatives are now being taken to bring these staff onto the government payroll.
450. In the case of FASE, - perhaps because the scale of funding required was too great to
permit a “go it alone mentality” to develop in the sector, perhaps because the DPs involved
had more rigorous processes for assessing the sustainability of spending plans – there was
a greater consciousness of the need to use the FASE common fund as a complement to
GoM funding through the budget. Thus, teacher salaries were never financed through FASE
and it was always understood that the Direct Support to Schools (ADE) grants would need
to be transferred to the Budget over time.
451. Notwithstanding the success of FASE, we would contend that most large-scale
Common Basket Funds are likely to generate the sort of perverse incentives, which have
been prominent in PROAGRI and PROSAUDE. (See Box 3). For these reasons, General or
Sector Budget Support is likely to be a less distortionary and more sustainable method of
providing external support to scale up service delivery and poverty-reducing spending. The
case of GBS in Mozambique provides an excellent example of how these processes may
work, even in quite difficult environments.
8.3. Recommendations
452. Our overall conclusion is that Budget Support has been fundamentally successful. It
has made possible a major expansion in education provision, whilst supporting economic
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 195
growth and macroeconomic stability and facilitating steady improvements in the quality of
public financial management and, to a lesser extent, in governance.. These are major
achievements, which fully justify the risks that have been taken in providing Budget
Support. These are still more impressive if one considers the relatively difficult context for
Budget Support, which has prevailed in Mozambique over the evaluation period.
453. The challenge for Mozambique and its Development Partners is to find ways of
achieving more with Budget Support in the future – in particular to find ways of making a
greater impact on poverty. With this as the overarching goal, we make a set of
recommendations for the joint consideration of the Mozambican authorities and its
Development Partners. These recommendations need further refinement and development,
based on inputs from the stakeholders on the ground. Their implementation will therefore
require a shared dialogue between the Mozambican Authorities and the G-19, in which
analysis and ideas are exchanged and a set of concrete actions is agreed and implemented,
aimed at increasing the effectiveness and impact of Budget Support in Mozambique.
454. We recommend that this dialogue, and the resulting action plan, should be clustered
around four objectives:
Renewing the commitment to the Budget Support process and the related aid
effectiveness agenda based upon realistic expectations not only of what it may
achieve but also of the investment of time and effort, which is required to make it
work.
Revitalising the Budget Support policy dialogue process, establishing a more
streamlined central process for assessing and reviewing performance, and creating
a stronger framework for sectoral policy analysis and dialogue, aimed at
stimulating good policy ideas and actions in the areas where they are currently
lacking.
Consolidating and deepening the progress achieved in education, in macroeconomic
management, in PFM reform and in the improvement of transparency and
accountability.
Focussing attention on poverty reduction through targeted programmes to reduce
the incidence of childhood malnutrition, increase the access of small farmers to
fertiliser and other inputs, and to create a conducive climate for business and
employment growth.
Renewing the commitment to the Budget Support process
455. Within this objective, the basic goal would be to first stop and then reverse the
disinvestment in the Budget Support process, identified by so many stakeholders in
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 196
Mozambique. “Half-hearted Budget Support” is not sustainable and should not be allowed
to continue. Among the G-19, those agencies unable or unwilling to invest in a full
commitment to Budget Support – for whatever reason – need to withdraw from the
process. Within the Government of Mozambique, there also needs to be a firm decision
about whether Budget Support is really wanted or not. A potential way forward to obtain a
renewal of the political commitment to Budget Support by the GoM and the G-19 members
would be a structured campaign at the technical level by the GoM and the G-19 to present
a more realistic message to their political leaders and supporting constituencies about
what Budget Support can and cannot achieve.
456. An obvious problem in the past has been that the expectations held for Budget Support
have been unrealistically high. Many DPs – in particular at headquarters level – expect to
be able to use Budget Support to exert “policy leverage” over partner governments so as to
change the governance context and achieve fast improvements in PFM and other reforms.
This has not happened in Mozambique nor in any other Budget Support recipient countries.
What Budget Support can do is firstly to provide funding for areas, which are clear shared
priorities and where tried and tested service delivery strategies exist – such as expansion
of education coverage and quality – and secondly to support and guide those institutional
reforms for which there is a government commitment.
457. A clear awareness also needs to be created regarding the commitments which Budget
Support entails both for GoM – in terms of transparency, exchange of information, an
openness to dialogue – and for the G-19 – in terms of a commitment to predictable
funding, to consistency in expectations, consistency in the treatment of problems, notably
cases of suspected violations of underlying principles, and to aid effectiveness principles.
The on-going revision of the MoU for Budget Support provides an excellent opportunity to
lay down common rules, regulations and procedures more carefully.
458. As a part of this process, there will also need to be a commitment – and a related set
of actions – to revitalise the processes of coordination across the G-19. The evaluation
team were unable to determine exactly why there has been deterioration in the quality of
coordination across the G-19. On the surface, there does not appear to be any particular
structural problem, nor any specific weakness with the frameworks that have been
established for coordination: the Troika Plus, the GBS secretariat, etc. However, it may be
that there are problems, which it was not possible to diagnose through the evaluation
process. An internal examination of the reasons for the deterioration in the effectiveness
of coordination is therefore recommended, based on a participatory process – potentially
managed by an external facilitator.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 197
Revitalising the Policy Dialogue process
459. Policy dialogue lies at the heart of Budget Support. It is essential that it should be
structured to be strategic (i.e. focused on issues of major importance to the national
development process), problem-solving (offering up solutions to policy blockages and
constraints) and efficient (as low in transaction costs as possible). At present, the
combined structure of central-level, sectoral and thematic groups within the dialogue
structure is not achieving these three objectives as effectively as it might. Again, we would
recommend an internally led, participatory diagnostic process, focused at two levels:
At the sectoral level, mechanisms need to be found to support effective processes
of research, debate and policy development. These mechanisms need to be
structured so as to strengthen policy development processes in the strategic areas,
where they appear to be weak – such as in the design and implementation of
agricultural policy. A structured, multi-annual programme of research and
evaluation guided through the Budget Support process should probably be created
as part of this mechanism. In addition, some facility could be created for bringing
into problematic sectors external facilitators, who might help to bring a greater
problem-focus into debate. In Tanzania, such a facility has been introduced
through the “Big Fast Results” initiative, led by the President’s Office and has
reportedly been helpful.
At the central level, efforts need to be made to streamline processes and increase
efficiency. The current Budget Support management framework works relatively
well and operates reasonably efficiently. However, attention should be given to
correcting the three areas of inefficiency in the current arrangements, which relate
to (i) the annual review framework created to mirror the PARP structure, which has
inadvertently introduced new transaction costs and over-complicated the dialogue
structures; (ii) the continued lack of precision in the definition of underlying
principles and the methods to be used to assess and monitor them; and (iii) the
size of the PAPs’ PAF and the overly ambitious targets which have been
established in the recent past for the indicators in the PAPs’ PAF.
Consolidating and deepening gains
460. Continued attention must be given to the success areas identified by this evaluation,
while adapting to the new challenges with regard to macroeconomic management, PFM
reform, governance and education. The major ‘success areas’ for Budget support in the
evaluation period were education, macroeconomic management, PFM reform, and, to a
lesser extent, governance.. There is every reason to expect that further progress can be
made in these important areas and it is therefore absolutely essential for the Government
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 198
and the G-19 to maintain the attention devoted to them in terms of policy dialogue,
budgetary resources and technical assistance and capacity building. At the same time, it
will be important in each of these areas to adapt the policy dialogue and, where relevant,
the focus of spending, to the new challenges which are emerging.
461. For macroeconomic management, it will mean greater attention to the framework for
investment selection and management and to the creation of the legal and institutional
framework for managing future revenues from the extractive industries. These are issues
where the lead advisory role must clearly be with the IMF in the context of the on-going
PSI. Hence the issue of the division of labour with the IMF and the appropriate roles for the
GBS PAF and for technical assistance from other sources needs also to be determined.
462. In relation to education, the challenge is firstly to strengthen the degree of attention
given to education quality issues and secondly to dedicate more attention to poor-
performing schools. As part of the FASE programme, the education sector is already
devoting more attention to learning achievement. There is work on going related to
developing a learning achievement baseline in primary education, which will be one of the
three indicators to be monitored at the level of the G19 GBS dialogue. This will also have
implications for the approach to teacher training and probably for textbook policy. The
other big challenge in the sector is to focus not just on improving average achievements
(average pupil-teacher ratios, average school performance, etc.) but also in finding ways of
reducing the variance around the mean by focusing attention more specifically on poor
performing schools and districts.
463. In relation to governance, the challenge is to convert gains in transparency into
tangible gains in accountability. The main improvements made over the evaluation period
relate to transparency – improvements in the availability of budgetary documents, and in
the scope and coverage of internal and external audit, and to a lesser extent to the
legislative framework for fighting corruption. Improvements in these areas need to be
continued and consolidated but efforts need also to be made to find more ways of using
information so as to exert accountability. We would recommend beginning with simple
measures – making internal and external audit reports more user-friendly and less
technical so that they can more readily be understood (and thus actioned) by public sector
managers, by Members of Parliament and by ordinary citizens. Mechanisms to follow up on
the implementation of recommendations also need to be tightened up by formalising and
documenting them and also by more active cooperation between the institutions of
accountability – the Inspeção Geral de Finanças (IGF), the Tribunal Administrativo (TA), the
Gabinete Central de Combate a Corrupção (GCCC) and the Comissão Parlamentar do Plano
e Orçamento (CPO). We recommend that each of these ideas should be considered for
inclusion within the work programmes for these areas.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 199
464. For PFM reform, the challenge is both to strengthen the design of the PFM Reform
Vision and to reinforce the institutional arrangements for the coordination of reforms and
for the provision of support to reforms. In relation to PFM reform design, there is a need on
the one hand to move from broad objectives to a more precise strategy, based on careful
attention to the sequencing of individual reforms, and on the other to broaden the focus of
reforms so as to address more fully the institutional requirements for controlling and
monitoring fiscal risks. With regard to coordination of reforms, there remains a need to
situate the coordination unit in the Ministry of Finance itself and to raise its profile and
relative level of authority so that it can be more effective in driving change across the PFM
system. In parallel, it is recommended that the G-19 should seek to strengthen its access
to high-level advisory support, so as to meet the challenges posed by the more complex,
‘second generation’ reforms.
Focussing on targeted programmes to reduce poverty
465. Finally, we recommend an expanded programme of Budget Support focused on specific
expenditure programmes and policies that might impact on poverty in the short to medium
term. The main shortcoming of Budget Support in the evaluation period has been the
failure to reduce income poverty. However, this is not just a failing of Budget Support but
of government policy as a whole. Most stakeholders now agree that PARPA I and PARPA II
focused too much on social sector spending and, while they were successful in improving
education and health outcomes, they failed to introduce the types of policy changes and
public expenditure programmes that could make a real difference to income poverty in the
short term. The PARP is certainly more focused on the income poverty problem but we
would argue that it is too vague to constitute a real strategy, based on concrete actions
and precise programmes.
466. Steady economic growth and fast increases in domestic revenue have created a degree
of fiscal space, which Mozambique has never experienced before. If projections for
continued economic and revenue growth are fulfilled, then domestic revenue should exceed
25% of GDP by 2018. This permits Mozambique to contemplate, from 2015 onwards, an
ambitious programme of public spending targeted on poverty reduction. Expanded provision
of Budget Support could allow such a programme to be quickly scaled up, without serious
risk to long-term fiscal sustainability. In addition to expanded social spending and
investment in physical infrastructure as in the past, we would recommend the design and
development of new programmes of spending and policy reform focused on three areas:
A programme of social transfers and educational inputs targeted to address
malnutrition: this should build on the work on Early Childhood Development, which
has been developed under the FASE programme but might also include school-
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 200
feeding programmes and cash transfers, focused on poor female-headed
households.
A programme to improve availability of fertilisers and inputs for small and
medium farmers: this should probably include measures to facilitate access to
credit in rural areas by simplifying the legal and institutional framework for credit
providers, actions to improve awareness of the importance of fertiliser and other
inputs, and to incentivise their marketing and distribution, and measures to
facilitate cross-border trade in agricultural inputs and in marketed output.
Improvements to rural transport infrastructure might also potentially form a part
of such a programme.
A programme of policy measures, conducted in close collaboration with the private
sector, aimed at improving the business environment and promoting employment
growth. This is likely to build on on-going work in this area being conducted in
conjunction with the World Bank.
467. The evaluation team looks forward to discussing these recommendations with
stakeholders from Government and the DPs, with a view to refining them and helping to
convert them into a set of precise, actionable measures. The challenge for Mozambique
and its Development Partners is to find ways of achieving more with Budget Support in the
future.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 201
Bibliography
ACS (2009), Avaliação da Implementação da Estratégia Anti-corrupção (2007-2009). Maputo, UTRESP, 27 de Julho de 2009.
Alfani, F et al. (2012) Poverty in Mozambique. New evidence from recent household surveys Policy research working paper 6217, The World Bank
APRM (2010), Country Review Report N0 11. Republico of Mozambique. Midrand, South Africa; OSISA (2009), Democracy and Political Participation, A review by AfriMAP and Open Society Foundation for Southern Africa.
Arndt C, et al (2011) Explaining poverty evolution. The case of Mozambique. UNU-WIDER Working Paper 2011/17
Brito, L (2010) O Sistema Eleitoral: Uma Dimensão Crítica da Representação Política em Moçambique.
Cabral, L (2009) Sector Budget Support in Practice – Mozambique Desk Study, Overseas Development Institute
Canadian Food Security Policy Group, Effective Aid for Small Farmers in Sub-Saharian Africa: Southern Civil Society Perspectives, January 2007.
Castel-Branco, N. (2008) Aid and Dependency and Development: A Question of Ownership? A Critical View. Working Paper 1, IESE,Maputo.
CEDSIF (2012) PFM Vision Document
CEDSIF (2010) Annual Report
CEDSIF (2011) Annual Report
CEDSIF (2012) Annual Report
CIA Fact book. Mozambique. Accessed 5/12/2013
CIP (2009) Relatório Anual de Controlo da Corrupção em Moçambique – 2008
CIP (2011) O Destino dos Nossos Impostos – A Conta Geral do Estado 2009 e Parecer do Tribunal Administrativo. Maputo, 2011.
Cumbi, A et al, 2012, Political economy of the health sector in Mozambique, MAP Consult
Cunguara, Benedito and Garret James. “Agricultural Sector in Mozambique: Situation analysis, constraints and opportunities for agricultural growth.” IFPRI, 2011.
Erskog, H and Rasmussen, P (2009) Analysis of External Funds Registered in the State Budget 2009. Draft (Unpublished)
European Commission, (2012) Monitoring of EU education and health expenditure in developing countries: Mozambique Case Study
Eurosis & MGA (2010), Diagnóstico Institucional do Tribunal Administrativo de Moçambique Maputo
FANR Directorate, SADC Secretariat (2008) Situation Analysis of Agricultural Research and Training in the SADC Region (Mozambique)
FAO/WFP (2010) (Crop and Food Security Assessment Mission to Mozambique. The Food and Agriculture Organization of the United Nations and the World Food Program, Rome
Friedrich-Ebert-Stiftung & MISA (2011) Barómetro Africano da Media. Moçambique
Hubert, D. (2012) Resources, Revenues and Social Development: Prospects and Challenges for the Extractive Industry in Mozambique, UNICEF, September 2012.
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 202
Hubert, D. (2013) Natural Gas from the Rovuma Basin: A ‘Game Changer for Mozambique?, UNICEF February 2013
IFPRI Discussion Paper 01229, The Supply of Inorganic Fertilizers to Smallholder Farmers in Mozambique”, December 2012.
INE, Agriculture and Livestock Census (2009-2010).
IMF (2006) Fourth Review Under the Three-Year Arrangement Under the Poverty Reduction and Growth Facility, Financing Assurances Review, IMF Country Report No. 6/254
IMF (2007) Republic of Mozambique: Sixth Review Under the Three- Year Arrangement Under the Poverty Reduction, IMF Country Report No. 07/262
IMF (2012) Republic of Mozambique: Fifth Review Under The Policy Support Instrument And Request For Modification Of Assessment Criteria, IMF Country Report No. 12/148
IMF (2013), Republic Of Mozambique, Staff Report For The 2013 Article Iv Consultation, Sixth Review Under The Policy Support Instrument, Request For A Three-Year Policy Support Instrument And Cancellation Of Current Policy Support Instrument, Country Report No. 13/200
IMF (2013) Republic of Mozambique: Fifth Review Under The Policy Support Instrument And Request For Modification Of Assessment Criteria, IMF Country Report No. 13/1
IMF (2014) Republic of Mozambique: First Review Under the Policy Support Instrument and Request for Modification of Assessment Criteria – Staff Report; Press Release IMF Country Report No. 14/20
Lawson, A et al (2006) Assessment Of Public Finance Management In Mozambique 2004/05 Based On PEFA Methodology, Final Report (Current Status Of PFM Systems & Processes, Overview Of Reforms And Perspectives For 2006), ODI / SAL & CALDEIRA,
Lawson, Andrew et al, (2008) Assessment of Public Finance Management & Procurement systems in the Mozambique Health Sector, based on an adaptation of the PEFA Methodology to the Health Sector, Final Report – English Version
Lawson, A et al (2009) Final Assessment Report 2008, Public Expenditure And Financial Accountability (PEFA) Assessment In Mozambique, SAL & CALDEIRA, FISCUS Limited / SAL & CALDIERA,
Martinez, S, Naudeau, S and Pereira, V (2012) The Promise of Pre-School in Africa: A Randomized Impact Evaluation of Early Childhood Development in Rural Mozambique World Bank/Save the Children
MB Consulting & CIPFA (2011). Avaliação de Despesa Pública e Responsabilidade Financeira (PEFA) em Moçambique. Relatório Final
Ministerio da Agricultura (2005) Trabalho de Inquerito Agricola. Maputo
Ministerio da Agricultura (2006) Trabalho de Inquerito Agricola. Maputo
Ministerio da Agricultura (2007) Trabalho de Inquerito Agricola. Maputo
Ministerio da Agricultura. (2008) Trabalho de Inquerito Agricola. Maputo
Ministerio da Agricultura (2011) Plano Estrategico de Desenvolvimento do Sector Agrario (PEDSA) Republica de Mocambique
Ministry of Agriculture (2011) Monitoring Agriculture Sector Performance, Growth and Poverty Trends in Mozambique, 2010 Annual Trends and Outlook Report, Directorate of Economics, Mozambique Strategic Analysis and Knowledge Support System (MozSAKSS) Mozambique
Ministry of Agriculture (2012) Monitoring Agriculture Sector Performance, Growth and Poverty Trends in Mozambique, 2011 Annual Trends and Outlook Report, Directorate of Economics, Mozambique Strategic Analysis and Knowledge Support System (MozSAKSS) Mozambique
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 203
Ministry of Agriculture (2010) Strategic Plan for Agricultural Development: PEDSA 2010-2019 Republic of Mozambique
MISAU, 2006, Avaliação Conjunta Annual do Desempenho do Sector de Saúde – 2005, ACA V
MISAU, 2008, National Plan for Human Resources Development 2008-2015
MISAU, 2008, Plano Estratégico do Sector Saúde: 2008 -2012
MISAU, 2009, Plano de Acção - Contributo para o desenvolvimento operacional do nível dos Sistemas de Gestão Financeira & Aquisiçõe
MISAU, 2009, Avaliação Conjunta Annual do Desempenho do Sector de Saúde – 2008, ACA VIII
MISAU, 2011, Relatório da Revisão do Sector de Saúde, 3rd Draft
MISAU, 2012, Relatório da Revisão do Sector de Saúde, Final
MISAU, 2012, Strategic Plan for Pharmaceutical Logistics, December 2012
MISAU, 2012, Plano de Acção para o Fortalecimento da Gestão Financeira do Sector de Saúde em Moçambique, Versão Final.
MISAU, 2013, Plano Estratégico do Sector Saúde: 2013 -2017, Pre Final Version
MISAU, 2013, Relatório de Execução Orçamental e Financeira, Sector da Saúde – 2012
MISAU, 2013, Avaliação Conjunta Annual do Desempenho do Sector de Saúde – 2012, ACA XII
Mogues, T and Benin, S (2012) Public Expenditures in Agriculture in Mozambique International Food Policy Research Institute
Molenaers, N, Capinskas, L and Jacobs, B (2010) Budget Support and Policy/Political Dialogue: Donor Practices in Handling (Political) Crises IOB/Universiteit Antwerpen
Mucavele, F (2009) The true contribution of Agriculture to the Economic Development of Mozambique, Paper presented at the Regional Stakeholders Policy Dialogue from 31st of August to the 4th of September 2009, in Maputo, Mozambique.
Ncube, S and Goris, W (2009) Trip report to Mozambique, July 2009 Hivos regional office Harare and Agri-ProFocus support office Arnhem
OECD-DAC (2006), Harmonising Donor Practise for Effective Aid Delivery: Volume 2; Budget Support, Sector Wide Approaches & Capacity Development in Public Finance Management, OECD-DAC; Paris
PAPs Secretariat / MPD Annual Review Aide Memoires 2004 to 2013
Pauw, K., Thurlow, J., Uaiene, R. and Mazunda, J. (2012) Agricultural Growth and Poverty Reduction in Mozambique: Technical Analysis in Support of the Comprehensive Africa Agriculture Development Program (CAADP). MozSAKSS Working Paper
PEFA (2010) Report on the Years 2007 – 2009 Draft version
Pessoa, M et al (2011) Mozambique: Follow-up of the Public Financial Management Reforms, International Monetary Fund
Pessoa, M et al (2012) Follow-Up of the Public Financial Management Reforms 2012
Pessoa, M et al (2012) Mozambique: Follow-up of the Public Financial Management Reforms 2012, International Monetary Fund
Republic of Mozambique, MISAU, Memorando de Entendimento
Republic of Mozambique, nd. CNCS, Memorando de Entendimento
Republic of Mozambiquen-European Community Country Strategy Paper 2008-2012
República de Moçambique, Tribunal Administrativo (2012) Demonstraçôes Financeiras Auditadas
Independent Evaluation of Budget Support to Mozambique, 2005 -2012 2014
Final Report; May 2014 P a g e | 204
Sida, Irish Aid and UK aid (2013) Summary Report of the Public Sector Governance Reform
Torbick, N.M. (2011) Mozambique: Agricultural Weather Risk Mapping: Final Report, Applied GeoSolutions LLC (for World Bank)
Transparency International (2012) Corruption Perception Index 2012. Berlin
Transparency International (2013) Global Corruption Barometer 2013. Berlin
Tribunal Administrativo (2010) Plano Estratégico do Tribunal Administrativo – PLACOR 2011-2014. Novembro 2010.
Umarji, M et al (2011) Final Assessment Report 2010, Public Expenditure And Financial Accountability (PEFA) Assessment In Mozambique MB Consulting / CIPFA,
Umarji, M. et al (2011) Avaliação de Despesa Pública e Responsabilidade Financeira (PEFA) em Moçambique 2010, MB Consulting, CIPFA
USAID (2010) Comparative Assessment of Decentralization in Africa: Mozambique In-country Assessment Report
Van den Boom, B (2011) Analysis of poverty in Mozambique. Household poverty status, child malnutrition and other indicators 1997, 2003, 2009
Visser-Valfrey, M and Bibi, M (2010) Sector Budget Support in Practice, Case Study: Health Sector in Mozambique, ODI/Mokoro
World Bank (2011) Mozambique - Analysis of Public Expenditure in Agriculture
Ximungo Consultores (2011) PSGR Case Study Mozambique: External Audit